The response of soil-atmosphere greenhouse gas exchange to changing plant litter inputs in terrestrial forest ecosystems

Various global change factors (e.g. elevated CO2 concentrations, nitrogen deposition, etc.) can alter the amount of litterfall in terrestrial forests, which could subsequently lead to changes in the physical, chemical, and biological properties of forest soils. Yet, there is hitherto a lack of consensus on the role of litter in governing the soil-atmosphere exchange of greenhouse gases (GHGs) in forest ecosystems, which can significantly affect the overall climatic cooling impacts of forests as a net carbon sink. In this study, we carried out a meta-analysis of over 250 field observations to determine the response of soil GHG fluxes to in situ litter manipulation in global forests. Our results showed that overall, litter addition enhanced soil CO2 emissions from terrestrial forests by 26%, while litter removal reduced soil CO2 emissions from these forests by 26%. The negative response of soil CO2 emissions to litter removal was stronger in the tropical forests (-33%) than in the subtropical (-27%) and temperate (-21%) forests, and was significantly correlated with mean annual temperature and precipitation. Moreover, litter removal was observed to enhance soil CH4 uptake in tropical (+24%) and temperate (+9%) forests, but not in subtropical forests. Litter removal reduced N2O emissions from forest soils by 20% on average, with this negative effect increasing with mean annual precipitation. The duration of litter removal experiment was negatively correlated with the response of soil CO2 emissions but had no influence on the response of soil CH4 and N2O fluxes. We found that plant litter supply could alter soil GHG fluxes in forests by modulating the microclimate as well as the labile and recalcitrant soil carbon pools. Our findings highlighted the importance of considering the effects of changing plant litter inputs on soil-atmosphere GHG fluxes in terrestrial forests and their spatio-temporal variability in biogeochemical models.