Around 1/3 of the Earth land surface is used for croplands. Their role in the carbon cycle is a crucial issue for scientists today. In the context of global warming, understanding the factors influencing carbon fluxes of agricultural soils and their components is essential for implementing efficient mitigation practices. The CO2 produced at the soil surface results from several respiratory processes making the evaluation of the existing methods complicated. Understanding the soil respiration sources and their dynamics are crucial issues to estimate the potential for carbon sequestration into soils via efficient cultural practices. Because of its major role in carbon loss over croplands, soil respiration modeling received much attention to quantify the fluxes (empirical modeling), to highlight the lack of knowledge and to guide researches (mechanistic modeling). In this study, empirical and semi-mechanistic models were carried out depending on how precise, generic or real the model should be. Using abiotic and biotic factors was essential to properly model respiration among five sites with different soil and climate. The Rh sensitivity to Ts and θs was adequate to obtain satisfying predictions over bare soils but the dependency of Rs on an indicator of the vegetation growth (GPP) was necessary to improve the predictions during crop periods. The empirical approach could not allow a good and reliable estimation of the contributions of the different components of Rs. Semi-mechanistic model was tested on 3 sites with various climatic and soil conditions. This approach allowed a good assessment on the heterotrophic and autotrophic contributions since it described more carefully the soil respiration and its underlying processes. Rh accounted for 63 % to 66 % of Rs for winter wheat culture whereas it accounted for 52 % to 56 % for a spring wheat rotation. Rs represented 33 % to 43 % of the total ecosystem respiration balance during winter wheat season and about 50 % for spring wheat. The semi-mechanistic model was developed to simulate the effects under different cultural practices as fertilization (manure) and tillage systems. It was concluded that carbon sequestration and carbon dioxide fluxes were more affected by soil organic matter inputs than by the tillage system itself.
