Predicting changes in soil organic carbon in mediterranean and alpine forests during the Kyoto Protocol commitment periods using the CENTURY model

Six Italian research sites, representative of Mediterranean and mountain forests and equipped with eddy covariance towers, were used in this study to test the performance of the CENTURY 4.5 model in predicting the dynamics of soil organic carbon (SOC) changes during the commitment periods (CP) of the Kyoto Protocol (2008-2012; 2013-2017). We show that changes in SOC stocks over short periods of time are difficult to detect, and explore the potential for models to be used for reporting SOC changes for forests that will remain forests, under Article 3.4 of the Kyoto Protocol. As the eddy covariance flux sites have been active for 10 yr on average, being initiated over the period between 1996 and 1998, the model was evaluated by comparing the modelled SOC stocks with those directly measured at each site in different years. Since long term series of observed values for soil carbon were not available, the validation of other model outputs such as net primary production (NPP) and soil nitrogen stocks, gives some confidence in long term simulations. Once the model performance was evaluated, two climate change scenarios, A1F1 (world markets-fossil fuel intensive) and B2 (local sustainability), were considered for prediction of C stock changes during the commitment periods of the Kyoto Protocol. In general, despite the need to consider the uncertainties in the direct measurements, at each site model fit with measured SOC stocks was good, with the simulated values within the standard deviation of the measurements. In this regard, the similarity between the SOC measured in 2008 and that predicted for the two forthcoming commitment periods points out the difficulty of detecting carbon stock changes by direct measurements, given the closeness in time to the present of the commitment periods. In any case, all sites show positive variations that are possibly related to the fertilization effects of increasing CO2 and to longer growing seasons, since no change in management occurred. Compared with the SOC measured in 2008, at the end of the second commitment period, the modelled SOC variations were smaller than 2% in the Mediterranean forests and comprised between 2% and 7% in the mountain forests. These variations, although small, indicate it might be possible to statistically detect differences after 10 yr in mountain forests with a reasonable number of samples. In conclusion, this work shows that since SOC stock changes are minimal within both CP, models can be effective tools for estimating future changes in SOC amounts, as an alternative to, or in support of, direct measurements when a short period of time is considered.