1.Atmospheric concentrations of carbon dioxide (CO2), methane (CH4) and nitrous oxide (N2O) are predicted to increase as a consequence of fossil fuel emissions and the impact on biosphere-atmosphere interactions. Forest ecosystems in general, and forest soils in particular, can be sinks or sources for CO2, CH4, and N2O. Environmental studies traditionally target soil temperature and moisture as the main predictors of soil greenhouse gas (GHG) flux from different ecosystems; however, these emissions are primarily biologically driven. Thus, little is known about the degree of regulation by soil biotic vs. abiotic factors on GHG emissions, particularly under predicted increase in global temperatures, and changes in intensity and frequency of precipitation events. 2.Here we measured net CO2, CH4 and N2O fluxes after 5 years of experimental warming (+3.4°C), and 2 years of ≈45% summer rainfall reduction, in two forest sites in a boreal-temperate ecotone under different habitat conditions (closed or open canopy) in Minnesota, USA. We evaluated the importance of microbial gene abundance and climo-edaphic factors (soil texture, canopy, seasonality, climate and soil physicochemical properties) driving GHG emissions. 3.We found that changes in CO2 fluxes were predominantly determined abiotically by temperature and moisture, after accounting for bacterial abundance. Methane fluxes on the other hand, were determined both abiotically, by gas diffusivity (via soil texture) and microbially, by methanotroph pmoA gene abundance, whereas, N2O emissions showed only a strong biotic regulation via ammonia-oxidizing bacteria amoA gene abundance. Warming did not significantly alter CO2 and CH4 fluxes after 5 years of manipulation, while N2O emissions were greater with warming under open canopy. 4.Our findings provide evidence that soil GHG emissions result from multiple direct and indirect interactions of microbial and abiotic drivers. Overall, this study highlights the need to include both microbial and climo-edaphic properties in predictive models in order to provide improved mechanistic understanding for the development of future mitigation strategies.