Article révisé par les pairs
Résumé : As the second largest area of contiguous tropical rainforest and second largest river basin in the world, the Congo Basin has a significant role to play in the global carbon (C) cycle. For the present day, it has been shown that a significant proportion of global terrestrial net primary productivity (NPP) is transferred laterally to the land-ocean aquatic continuum (LOAC) as dissolved CO2, dissolved organic carbon (DOC), and particulate organic carbon (POC). Whilst the importance of LOAC fluxes in the Congo Basin has been demonstrated for the present day, it is not known to what extent these fluxes have been perturbed historically, how they are likely to change under future climate change and land use scenarios, and in turn what impact these changes might have on the overall C cycle of the basin. Here we apply the ORCHILEAK model to the Congo Basin and estimate that 4% of terrestrial NPP (NPPD5800-166 TgC yr-1) is currently exported from soils and vegetation to inland waters. Further, our results suggest that aquatic C fluxes may have undergone considerable perturbation since 1861 to the present day, with aquatic CO2 evasion and C export to the coast increasing by 26% (186-41 to 235-54 TgC yr-1) and 25% (12-3 to 15-4 TgC yr-1), respectively, largely because of rising atmospheric CO2 concentrations. Moreover, under climate scenario RCP6.0 we predict that this perturbation could continue; over the full simulation period (1861-2099), we estimate that aquatic CO2 evasion and C export to the coast could increase by 79% and 67 %, respectively. Finally, we show that the proportion of terrestrial NPP lost to the LOAC could increase from approximately 3% to 5% from 1861-2099 as a result of increasing atmospheric CO2 concentrations and climate change. However, our future projections of the Congo Basin C fluxes in particular need to be interpreted with some caution due to model limitations. We discuss these limitations, including the wider challenges associated with applying the current generation of land surface models which ignore nutrient dynamics to make future projections of the tropical C cycle, along with potential next steps.