Microbial feedbacks to climate change on local to global scales

Background/Question/Methods

Feedbacks between climate warming and carbon cycling on land are a major source of uncertainty in future climate projections. Soils play a key role in this uncertainty because they are expected to lose carbon with warming due to increased decomposition but may also gain carbon due to increases in plant primary production. Biogeochemical models are used to analyze the balance between these two factors, but current models omit key microbial processes. Improving the accuracy of model predictions requires answers to four main questions: 1) How variable are current model predictions, and what causes the variation? 2) How accurate are current models when compared to observational data? 3) What are the consequences of changing models to account for advances in microbial ecology? 4) How do we scale these microbial processes? I synthesized models and data at a range of scales to address these questions.

Results/Conclusions

Current biogeochemical models diverge widely in their soil carbon projections and differ qualitatively in behavior compared to microbial models. For example, current models simulate global carbon stocks of 510 to 3040 Pg C compared to observations of 890 to 1660 Pg C. Going forward, these models project changes in global soil carbon ranging from losses of 72 Pg C to gains of 253 Pg C over the 21^st century. Models with large gains generally simulate large increases in high-latitude net primary production. New models that account for microbial physiology and enzyme kinetics can explain up to half of the spatial variation in contemporary soil carbon stocks, roughly double the variance explained by the best conventional models. Still, these microbial models are relatively unproven. They project a wide range of soil carbon responses to 21^st century global change, and they are surprisingly insensitive to changes in carbon inputs. New analyses are required to parameterize microbial models and scale up microbial physiology. Small-scale models that account for microbial diversity and functional traits are promising tools for addressing this challenge.