The boreal forest and tundra biomes are also very poorly represented in terms of eCO2 research (Fig. 2a). Estimates suggest that together 540–1700 Gt of C is stored in the soils and living biomass of these biomes (UNEP-WCMC, 2008 and Tarnocai et al., 2009) (see Supplementary data S1). Most C (ca. 85%) in the boreal forest biome is stored in soil (Malhi et al., 1999) and understanding the response of this immense carbon reserve to combined global changes, including eCO2, remains a research priority. It is uncertain whether increased C sequestration will occur with eCO2 conditions and under a warming

atmosphere. However, we need to establish if the addition of new carbon, particularly with warmer conditions, is likely to prime the release of old carbon from these soil stores BLU9931 mouse (Freeman et al., 2004 and van Groenigen et al., 2014), thereby positively feeding back on eCO2. From our synthesis we conclude that a global strategy for eCO2 research needs to be completed. Outstanding needs include

accounting for remaining uncertainty in the effects of eCO2 on plant productivity and soil C Z VAD FMK storage. Such information is essential in order to effectively predict global C dynamics under a future eCO2 climate, particularly in the most understudied ecosystems with the greatest potential influence on C dynamics globally. At a global scale, these are the highly productive forests of the tropics (Pan et al., 2011) and the soils of tundra and boreal regions (Tarnocai et al., 2009), both of which have been largely overlooked by long-term eCO2 research programs. Long term eCO2 experimentation in these areas would support integrated modeling with improved resolution for these biomes, in order to integrate plant Gemcitabine and soil processes at the global scale. To be effective, this research would need be coordinated and follow standardized protocols for plant productivity assessments and soil C fluxes. This could be integrated with existing global carbon dynamics studies that have standardized methodologies for

C dynamics monitoring, such as the Global Ecosystems Monitoring Network (GEM) which uses a network of 1 ha forest plots (Marthews et al., 2012). A network of spatially smaller eCO2 experiments could be embedded to build on existing knowledge and expertise. Such an approach would deliver a thorough account of above and below ground fluxes in both plant productivity and soil carbon in response to eCO2. By standardizing measurements and instrumentation, direct comparisons could be made between a range of forest plant communities, thereby allowing the spatial and temporal limits of the CO2 fertilization effect to be quantified according to climate, habitat type and disturbance history, within major biomes for C sink activity. Importantly the new generation of eCO2 experiments needs to be designed to have a low carbon footprint, possibly utilizing CO2 “wastes” and local resources (e.g.