Improving seasonal carbon dynamics in contrasting Mediterranean and Alpine grasslands

Grasslands cover a substantial part of the global land area (~40%) and store about one third of the terrestrial carbon stock. These ecosystems and their significant carbon stocks are very susceptible to climate change and are often extensively managed for human use. Nevertheless, land surface models consistently fail to predict carbon fluxes in grasslands accurately, which is probably due to the lack of a good phenology module. Grassland vulnerability to climate change in combination with their large carbon stocks calls for an improved representation of grasslands in land surface models (LSMs) to accurately predict their fate under a changing climate.  Here, we use data from two ecosystem manipulation experiments (MaNiP, Nitrogen and Phosphorus fertilisation in a Mediterranean tree-grass ecosystem and ClimGrass, drought, warming and elevated CO2 in a montane grassland) to improve the representation of carbon dynamics in  the LSM QUINCY. We built a novel turnover and growth model representation for both vegetative and reproductive plant pools in QUINCY based on ecologically realistic temperature and moisture controls as well as plant life history strategies. We show that the modified model can capture seasonality of productivity both in the seasonally cold and seasonally dry systems, under ambient and experimental conditions. This generalised model built upon manipulative experiments will improve global grassland productivity predictions.