Soil organic carbon changes at the plot level in hillside systems

To determine changes in soil organic carbon (SOC) stocks at the local or regional scales, it is necessary to improve our understanding of the carbon (C) cycle and to take inventory of this element in different ecosystems and land uses. This study was conducted in three regions of the Sierra Norte of Oaxaca to estimate SOC changes in plots under agricultural, forest and grassland systems. In 2000 and in 2004 SOC was measured directly in soil samples collected in the field to detect changes in SOC, which were later compared with simulated SOC changes performed with the RothC-26.3 model. The RothC was executed with SOCinitial (SOC 2000) measured at each sampling point (*CIPUN) within one plot and with the SOCinitial averages of the sampling points for each plot (*CIPAR). The measured and simulated SOC changes were compared under three grouping criteria: individual plot (27), land use system (3) and without consideration to system or plot (367). The direct SOC measurements in these systems showed changes over the four-year period ranging from −1 to 25 Mg C mass ha−1 accumulated in the top 30 cm of soil, in general, higher than those reported in the literature. Among the possible causes, the following might be included: overestimation of C percentages of the measurements taken in 2004 and the 25–77% CV from natural causes or management. The best performance of the RothC by plot was achieved when the model was executed using *CIPUN (R2 of 0.90 and EF (efficiency) of 0.67, but when the performance of the model was analyzed by land use system, the cropping systems obtained an R2 of 0.76 and EF of 0.56. Finally, when no consideration was made to plot or system the RothC simulations indicated an underestimation of 33% with respect to the measured SOC, and the trend between simulated and measured SOC had a correlation coefficient of 0.84.