Agricultural land use changes are expected to modify the carbon (C), nitrogen (N) and phosphorus (P) stocks compared to the native ecosystems they replace and result in changes in greenhouse gas (GHG) fluxes. To quantify these effects, we measured C-N-P stocks in four land cover classes (cropland, grassland, native shrubland and invaded shrubland) together with their ecosystem-level CO2, CH4 and N2O fluxes for one year using the closed-chamber technique, in three locations on Chiloé Island, southern Chile. As a reference, given that most of the original ecosystems were derived from forest clearance, we monitored CO2 fluxes in a native forest site in the same area using the eddy covariance technique, and used published data of its C-N-P stocks and CH4 and N2O fluxes. We found that compared to the forest, the other land cover classes represented a 60 % loss of total ecosystem C, which was explained mainly by the loss of 98 % in above ground biomass and a 42 % reduction in soil C, but with a relatively unchanged N stock and an increase in soil P. While croplands were net GHG emitters (901 g CO2-eq m-2 year-1), grasslands, native shrublands and invaded shrublands were net sinks (−1350, −1154 and −2088 g CO2-eq m-2 year-1, respectively). The invaded shrublands (invaded by the N-fixing species Ulex europaeus and Cytisus scoparius) showed higher aboveground and root biomass compared to the other land cover classes, but somewhat lower C-N-P soil stocks, possibly because of lower litter decomposition, reduced root turnover/rhizodeposition and/or greater dissolved elemental losses, as the invasion occurred on degraded lands. Although the invasive shrublands acted as a stronger GHG sink, driven by their higher CO2 fixation, this might change in the longer term through the additional accumulation of fixed N, with a potential increase in N2O emissions.
