Agricultural expansion in the Brazilian Cerrado ecoregion has been causing extensive land use and land cover changes (LULCC), drastically shifting the carbon cycle dynamics of the affected ecosystems. However, accurate in situ observations of the net ecosystem exchange of carbon dioxide (NEE) from wooded Cerrado (Cerrado sensu stricto) as well as from post-conversion agricultural landscapes are lacking, with the limited amount of impact assessments in the literature being primarily based on remotely sensed data. This study presents a multi-annual time series of temporal high-resolution eddy covariance carbon dioxide fluxes, measured on the border between a wooded Cerrado and a post-conversion agricultural area, primarily used as a pasture, in southeastern Brazil. We investigated multiple setups of NEE partitioning methods to separate NEE into its components gross primary production (GPP) and total ecosystem respiration (TER). We combined these component partitioning models with source area partitioning methods to estimate component fluxes for the two contrasting ecosystems within the tower footprint. Model results were compared against remotely sensed vegetation indices and flux data from similar ecosystems. We found that converting native wooded Cerrado to a pasture-dominated agricultural area decreased the landscape’s NEE carbon (NEE-C) uptake by up to 494 g m-2 yr-1 (73 %). The wooded Cerrado had an annual cumulative NEE-C of -639 ± 20 g m-2 yr-1 and -673 ± 19 g m-2 yr-1 in 2019 and 2020, respectively. In comparison, the pasture had lower annual cumulative NEE-C of -146 ± 39 g m-2 yr-1 and -179 ± 38 g m-2 yr-1 in the same years. The pasture exhibited lower light use efficiency (LUE) and NEE-C uptake in the dry season, resulting in lower annual NEE-C uptake. Additionally, the pasture showed greater sensitivity to precipitation changes, leading to higher seasonal variations in carbon dioxide fluxes.
