Summary paragraph 34Plants acquire carbon through photosynthesis to sustain biomass production, autotrophic 35 respiration, and production of non-structural compounds for multiple purposes 1 . The fraction 36 of photosynthetic production used for biomass production, the biomass production 37 efficiency 2 , is a key determinant of the conversion of solar energy to biomass. In forest 38 ecosystems, biomass production efficiency was suggested to be related to site fertility 2 . Here 39 we present a global database of biomass production efficiency from 131 sites compiled from 40 individual studies using harvest, biometric, eddy covariance, or process-based model 41 estimates of production -dominated, however, by data from Europe and North America. We 42show that instead of site fertility, ecosystem management is the key factor that controls 43 biomass production efficiency in terrestrial ecosystems. In addition, in natural forests, 44 grasslands, tundra, boreal peatlands and marshes biomass production efficiency is 45 independent of vegetation, environmental and climatic drivers. This similarity of biomass 46 production efficiency across natural ecosystem types suggests that the ratio of biomass 47 production to gross primary productivity is constant across natural ecosystems. We suggest 48 that plant adaptation results in similar growth efficiency in high and low fertility natural 49 systems, but that nutrient influxes under managed conditions favour a shift to carbon 50 investment from the belowground flux of non-structural compounds to aboveground biomass. 51 52 53 Main text 54The fraction of gross primary production (GPP) used for biomass production (BP) of 55 terrestrial ecosystems has recently been coined biomass production efficiency (BPE) 2 . BPE is 56 typically used as a proxy for the carbon-use efficiency or NPP-to-GPP ratio, where NPP refers 57 to net primary production i.e. BP plus the production of non-structural organic compounds 1 . 58 4 Current knowledge about BPE is mainly derived from research on forests. Earlier work 59 reported BPE to be conservative across forests 3 , whereas more recent syntheses suggest high 60 inter-site variability 2,4 . The variation in BPE was first attributed to vegetation properties 61 (forest age) and climate variables 4 . More recently, it was shown that forest BPE in a range of 62 natural and managed sites was correlated with site fertility, with management as a secondary 63 BPE driver 2 . 64Fertility and management are strongly correlated as management enhances 65 productivity by increasing plant-available resources, including nutrients. For instance, 66 fertilization of grasslands directly increases the ecosystem nutrient stock, whereas forest 67 thinning indirectly increases nutrient availability at the tree level by reducing plant-plant 68 competition. In addition, fertile sites are more likely than infertile sites to be managed. 69Atmospheric deposition of nutrients, especially nitrogen (N), might further complicate the 70 relationship between BPE, fertility and ...

Multiple lines of evidence 1,2 have established the existence of a land carbon sink in the northern hemisphere, but its trend remains poorly understood 3,4 . Here we show from measurements of the atmospheric CO2 gradient between the stations of Mauna Loa (north) and of South Pole (south) since 1958 that the northern land sink remained rather stable between the 1960s and the late 1980s, but increased by 0.5 ± 0.4 Pg C yr -1 during the 1990s and further by 0.6 ± 0.5 Pg C yr -1 during the 2000s. The first increase of the northern land sink in the 1990s accounts for 65% of the increase in the global land flux during that period. The subsequent increase in the 2000s is larger than the increase of the global land flux, suggesting a coincident decrease of carbon uptake in southern regions. The decadal change in the northern land sink between the 1960s and the 1990s can be explained with a combination of rising atmospheric CO2, climate variability, and changes in land-cover as represented in an ensemble of terrestrial carbon cycle models 4 . The increase during the 2000s is however underestimated by all terrestrial models. Reducing uncertainties involve better consideration of changes in drivers such as land use change, diffuse light, and nitrogen deposition as well as denser observations to clarify the importance of different regions.Fossil fuel and land-use change emissions of CO2 into the atmosphere increased by a factor of three between 1960 and 2016, with a fast growth rate in the 1980s, a slower growth rate in the 1990s and a re-acceleration 5 in the 2000s. The global land and ocean carbon sinks increased proportionally with growing emissions 6 but their location and trends are incompletely understood. Northern Hemisphere (NH) lands make a dominant contribution to the global land carbon sink 1-4,7-8 . In the NH mid and high latitudes, vegetation greenness increased in the last 30 years 9 , and the seasonal amplitude of CO2 increased by 50% in the last 50 years 10 , suggesting an enhancement of ecosystem production.However, these observations are not proofs that the NH net carbon sink is increasing because of possible upward trends in soil respiration 11 and land-use emissions compensating for increased production.To gain insights on the long-term trend in the northern land sink over the last 50 years, we use the inter-hemispheric gradient of atmospheric CO2 (IG) defined as the observed difference of atmospheric CO2 between the Mauna-Loa (MLO) station located at 19°N, and South Pole (SPO).Both MLO and SPO record CO2 growth rates representative of the means in their respective hemisphere 12 . Here we examine the relationship between IG and fossil fuel and cement CO2 emissions (F) between 1958 and 2016, and recent changes during the 2000s, a period marked by the acceleration of global CO2 emissions, mainly from East and South Asia regions [13][14] .From 1958 to 2016 the IG grew proportional to F (Fig. 1; Fig. ED1) with a Pearson correlation coefficient (r) of 0.97 (p <0.01) and a mean regression slope 15 of 0.44± 0.01...

Concentrations of atmospheric carbon dioxide (CO2) have continued to increase whereas atmospheric deposition of sulphur and nitrogen has declined in Europe and the USA during recent decades. Using time series of flux observations from 23 forests distributed throughout Europe and the USA, and generalised mixed models, we found that forest-level net ecosystem production and gross primary production have increased by 1% annually from 1995 to 2011. Statistical models indicated that increasing atmospheric CO2 was the most important factor driving the increasing strength of carbon sinks in these forests. We also found that the reduction of sulphur deposition in Europe and the USA lead to higher recovery in ecosystem respiration than in gross primary production, thus limiting the increase of carbon sequestration. By contrast, trends in climate and nitrogen deposition did not significantly contribute to changing carbon fluxes during the studied period. Our findings support the hypothesis of a general CO2-fertilization effect on vegetation growth and suggest that, so far unknown, sulphur deposition plays a significant role in the carbon balance of forests in industrialized regions. Our results show the need to include the effects of changing atmospheric composition, beyond CO2, to assess future dynamics of carbon-climate feedbacks not currently considered in earth system/climate modelling.