Most of the Earth System Models (ESMs) rely on empirical functions for decomposition of litter with multiple soil carbon pools decaying at different rates to estimate R_eco variations and its partitioning into autotrophic (R_a) and heterotrophic respiration (R_h) in relation to variation in temperature and soil water content. However, microbially-mediated litter decomposition schemes are very scarce in ESMs. Microbial enzymatic processes are integral parts in litter as well as soil organic matter (SOM) decomposition. Here we developed a mechanistic model comprising of multiple hydro-biogeochemical modules to explicitly incorporate microbially-enzymatic litter decomposition and decomposition of SOM for estimating regional-scale R_a, R_h and R_eco. Modeled annual mean R_eco values are found varying from 1000 to 8000 kg C ha^-1 yr^-1 in 2000-2013 within the boreal forest covered sub-basins of the Athabasca River Basin (ARB), Canada. While, for the 2000-2013 period, the annual mean R_a and R_h are varying within 800-6000 kg C ha^-1 yr^-1 and 500-3500 kg C ha^-1 yr^-1, respectively. R_a generally dominates R_eco with nearly 30-80% contribution in most of the sub-basins in ARB. The model estimates corroborate well with the site-scale and satellite-based estimates reported at similar land use and climatic regions. Modeling the partitioning of R_eco to R_a, and R_h are critical to understanding future climate change feedbacks and to help reduce uncertainties in ESMs in the boreal and subarctic regions.