At present, the terrestrial biosphere is mitigating anthropogenic climate change by acting as a carbon (C) sink, compensating about 30% of global CO 2 emissions from fossil and land-use sources 2 . In contrast, 44-73% of global nitrous oxide (N 2 O) emissions 3,4 and 24-43% of global methane (CH 4 ) emissions 5 , both potent GHGs, originate from land ecosystems and partly offset the cooling effect of C uptake by the land. Terrestrial N 2 O and CH 4 emissions, henceforth termed eN 2 O and eCH 4 , are enhanced in a warm climate 6,7 and under high atmospheric CO 2 concentrations (cCO 2 ; ref. 8). The associated feedback loop amplifies anthropogenic climate change and is reflected in palaeo records on glacial-interglacial and centennial timescales 9,10 . However, despite its potential importance 6 there is yet a lack of studies investigating combined multiple GHG feedbacks between terrestrial ecosystems and climate.The strength of feedbacks between land and climate is determined by the sensitivity of the forcing agents (here: eN 2 O; eCH 4 ; terrestrial C storage, C; and Albedo change) to the drivers (climate and cCO 2 ), and the radiative efficiency of the respective forcing agent. Earlier quantifications of terrestrial GHG feedbacks have relied on observational data and land-only models to derive the sensitivities, multiplied by the radiative efficiency 6,7,9-11 . Here, we assess multiple feedbacks from terrestrial ecosystems in a coupled Earth system model of intermediate complexity and follow a quantification framework commonly applied to measure the strength of physical climate feedbacks 1,12 ( Fig. 1 and Methods). Applying future scenarios of N-deposition and Nfertilizer application in agriculture allows us to assess their impact on eN 2 O and related feedbacks.We start the discussion by exploring to which extent a processbased land biosphere model is able to reproduce the observationbased evolution of atmospheric N 2 O and CH 4 concentrations (cN 2 O, cCH 4 ) over the industrial period. Addressing the historical atmospheric GHG budgets serves as a test for the sensitivity of simulated GHG emissions to the combination of climate, cCO 2 and external forcings. LPX-Bern 13-18 is applied here to simulate the coupled cycling of carbon and nitrogen and the emissions of GHGs from agricultural and natural land and from peat. Sitescale evaluations of this model have been presented earlier 7,[15][16][17][18][19][20] . For this test, we force LPX-Bern with observational data for climate 21 , cCO 2 (ref. 22), Nr (N deposition 23 plus mineral N fertilizer inputs 24 ) and anthropogenic land-use area change and combine simulated emissions with independent emission data from remaining sources to assess atmospheric budgets (see Methods and Supplementary Fig. S1).For eN 2 O, we confirm earlier results 24 showing that the simulated emission increase in the second half of the twentieth century matches measured concentrations (Fig. 2a). Experiments with incomplete driving factors perform worse at reproducing the observed rate of i...