CHANGE IN THE ORGANIC-CARBON CONTENT OF S OIL IN A LONG-TERM CONTINUOUS NONIRRIGATED MAIZE CROP

In 2 long-term experiments located on loamy sandy soils in southwest F rance, change in soil carbon status was measured over a 22- and 25-yr period respectively. Soil cropping was as follows: continuous maize cr op with stalks returned to the soil (TR), continuous maize crop with s talks removed (TE), continuous maize crop including a winter annual co ver as rye-grass (RG) and rotation involving temporary grassland for 3 yr followed by a maize crop for 4 yr (PT). Whatever the treatment app lied, the soil organic carbon content decreased during the course of e xperiment In the plots under continuous maize cropping with returned s talks, soil carbon concentration decreased from 8.30 to 6.92 g/kg (32 to 26 t organic C ha(-1)) over 22 yr and from 15.50 to 9.65 g/kg (57 t o 36 t C ha(-1)) over 25 yr respectively in both experiments. No marke d change in carbon balance due to the various organic matter treatment s occurred during the course of experiment. The amount of stored C dec reased by 3 and 4 t organic C ha(-1) respectively in both experiments as a result of continuous removal of plant residues. On the other hand , using a winter cover crop of rye-grass or the establishment of a tem porary grassland resulted in an increase of C storage ranging from 2 t o 3 t ha(-1) or from 2 to 4 t ha(-1) organic C. On the contrary, intro ducing a cultivated fallow in 1980 resulted in a marked decay of organ ic matter averaging 7 t ha(-1) C for both experiments in 1988 and in 1 991 respectively. An asymptotic exponential model led to the predictio n that the ploughed soil layer would contain 23 and 31 t ha(-1) organi c C (6.1 and 8.4 g/kg C content) at the equilibrium stage. Using the H enin-Dupuis relationship, which involves parameters based on field mea surements (k(1) = 5.6 and 6.5%; k(2) = 1.16 and 1.63%), calculated equ ilibrium values of soil carbon level content are much lower. Taking in to account data from experiments carried out in other countries, such low values are questionable. The unreliability of the Henin-Dupuis (19 45) model could be attributable to ifs single-compartmental conception and to an unsatisfactory approach using the k(1) and k(2) parameters calculation. Indeed, such a calculation does not take into account the marked change in soil mineralization due to the various cultural prac tices or different organic matter fluxes entering the soil in the fiel d experiment that result in a systematic underestimation of predicted equilibrium values.