Local alterations of land uses by policy, planning, and management decisions have global implications for coupled biogeochemical cycles. Quantification and prediction of impacts of land-use changes on carbon (C), nitrogen (N), and water (H2O) cycles are of great significance, in particular to the Mediterranean ecosystems that are already vulnerable to climate change. The present study was aimed at empirically modeling the four response variables of soil carbon (SC), nitrogen (SN) contents, carbon dioxide (CO2), and H2O effluxes as a function of the 10 predictors of land use type (forest, grassland, cropland, and their degraded states), soil organic matter, soil moisture, silt, clay and sand fractions, pH, electrical conductivity, soil microorganisms, and soil temperature. Our results showed that soil respiration rate was highest for cropland and lowest for forest (p = 0.002). Land use type was found to be the primary control and significantly related linearly to SC, SN, and soil CO2 efflux and non-linearly to all the responses. Goodness-of-fit and predictive power of the best-fit multiple non-linear regression (MNLR) models varied between 80.8% for soil CO2 efflux and 99.9% for SC, and between 67.4% for soil CO2 efflux and 99.1% for SN, respectively.