The evaluation of the potentialities of carbon storage in the cultivated agro-systems is on the agenda. In this respect, the aim of our research is to evaluate the stocks of C in the soil and show the mechanisms concerned, under the action of farming practices in a ferric Acrisol of Burkina Faso.

This study is based on long-term agronomic experiments which take into account the time aspect for understanding, quantifying and modelling the evolution of C stocks as a result of ordinary farming practices that can be optimised, such as mineral fertilisation, organic matter input (straw, compost, manure) and soil cultivation (scraping and ploughing). Measurements and analyses are carried out in the field and in the laboratory: (1) in the field with the soil (morphological characterization, bulk density) and with the crops (outputs), (2) in the laboratory with the organic matter input (biochemical composition) and with soil sample (dosage of C and N, diffraction and particle-size fractionation).

The results corroborate a decrease of the C stocks due to farming, in comparison with the initial herbaceous fallow, after 10, 20, and 40 years of continuous farming including ploughing of 42, 44 and 68%. The C storage potential level on this type of soil measured on the non-deteriorated herbaceous fallow is 20.6 Mg ha^-1 in the 0-20 cm horizon. The most powerful practices tested make it possible to reach 45 to 90% of this potential stock, according to the duration of the experimentation. The yearly net mineralization coefficients are high, ranking from 0.035 to 0.038. The isohumic coefficients are 18 and 22 respectively for doses of 40 and 5 Mg DM ha^-1 2ans^-1 of manure. The modelisation of C stock with the RothC model (0-20 cm horizon) brings us to admit the existence of an unsuspected cause of loss or transfer from manure of about 67% which does not act on the improvement of C stock of the soil of the concerned horizon. This has seemingly something to do with the role of the soil's macrofauna. The importance of the 0-20 µm fraction and soluble is shown in the particle-size fractionation. The latter is the highest (30% of the total C) without any organic matter input. The fine fractions are relatively stable whereas important variations are reported for the coarse fractions in case there is no input of organic matter. When there is a manure input together with nitrogen fertilizer, this situation is favourable for preferential storage of C in the fine fractions, thus showing the advantage of this practice in the stabilization of the SOM. The definition of a critical value of C in the soil, between 6 and 7 mg C g^-1 is confirmed by the convergent results of two approaches: the analysis of sorghum outputs with the “envelope curve” method and the definition of a C sill value for the sustainability of the system (Feller 1995) which show 6 and 6.8 respectively. The outputs decrease below 6 mg C g^-1 and stabilize above this value.

Increasing the storage of C in the soil is possible with the current farming practices but their optimisation must be achieved. This study opens onto possible actions and concludes on the necessity to take into account the macrofaunic and microbial activities of the given soil in interaction with the organic matter brought for a better control of the organic flows and evaluation.

Keywords: carbon stock, long-term experiments, ferric Acrisol, Burkina, farming practices, organic matter, manure, compost, particle-size fractionation, modelling.