The sources and transformation mechanisms of HCO3- in outlet water bodies of typical karst watersheds are crucial for conducting karst carbon sink accounting and assessing carbon sequestration capacity. This study was conducted at the Yaji Experimental Site in Guilin, Guangxi, from the perspective of earth system science, employing a combination of high-resolution monitoring and monthly sampling techniques. Continuous measurements temperature (T), electrical conductivity (EC), and CO2 were conducted at soil depths of 10 cm, 40 cm, and 90 cm. Simultaneously, soil water, air, and spring water samples were collected for the analysis of pH, CO2, δ13CDIC, HCO3-, and other hydrochemical parameters. By coupling mathematical models with hydrogeochemical techniques, we utilized data-driven linear mixing models and hydrochemical analyses to systematically interpret the sources, influencing factors, and migration characteristics of HCO3- within the soil-groundwater system. The proportion of HCO3- derived from soil CO2 was found to be 80.00% in soil water and 70.06% at the karst spring outlet (S31), indicating that soil CO2 is the dominant source of HCO3- in the soil-groundwater system. Soil CO2 levels were primarily influenced by meteorological conditions, soil temperature, soil EC, and biological activity. During the rainy season, increased soil CO2 and enhanced karst dissolution resulted in simultaneous rises in Ca2+ and HCO3- concentrations in S31. Based on the hydrochemical facies, calcite saturation index, and CO2 degassing characteristics, it was found that the CO2-HCO3--CaCO3 system in the soil-groundwater continuum reaches a dynamic equilibrium. Overall, the dynamics of HCO3- in the karst small-watershed soil-groundwater system are governed by the coupled processes of “soil CO2 generation, water–rock reaction, and hydrological transport”. This system exhibits a migration-transformation pattern characterized by “rapid generation in upper layers, enrichment in the middle layers, and stable output at the spring outlet”. Moreover, reprecipitation at the karst spring outlet acts as a mechanism for reducing HCO3- concentrations. The findings of this study are expected to provide scientific guidance for quantifying karst carbon sink fluxes and assessing carbon sequestration potential.