The diversity of soil adsorbents for arsenic (As) and the often-overlooked influence of manganese (Mn) on As(III) oxidation impose challenges in predicting As adsorption in soils. This study uses Mössbauer spectroscopy, X-ray diffraction of oriented clay, and batch experiments to develop a kinetic coupled multi-surface complexation model that characterizes As adsorbents in natural soils and quantifies their contributions to As adsorption. The model integrates dynamic adsorption behaviors and Mn-oxide interactions with unified thermodynamic and kinetic parameters. The results indicate that As adsorption is governed by five primary adsorbents: poorly crystalline Fe oxides, well crystalline Fe oxides, Fe-rich clay, Fe-depletion clay, and organic carbon (OC). Fe oxides dominate As adsorption at low As concentrations. However, at higher As concentrations, soils from carbonate strata, with higher content of Fe-rich clay, exhibit stronger As adsorption capabilities than soils from Quaternary sediment strata. The enrichment in Fe-rich clay can enhance the resistance of adsorbed As to reduction processes affecting Fe oxides. Additionally, extensive redox cycles in paddy fields increase OC levels, enhancing their As adsorption compared to upland fields. This model framework provides novel insights into the intricate dynamics of As within soils and a versatile tool for predicting As adsorption across diverse soils.