| Popis: |
Context and background Forest ecosystems play a key role in the regulation of carbon cycle. The sustainability of This ecosystemic service is a necessity in the context of global changes. The spatial distribution of carbon stock is a useful tool of decisionmaking. Which contribute to improve forest management practices. Goal and Objectives: The aim was to model the spatial distribution of the aboveground carbon stock in the Goungoun Forest Reserve and its surrounding lands in northern Benin. Methodology: Systematic stratified sampling was used for the forest inventory. One hundred and fifty circular sampling plots with a radius of 18 m were inventoried. In each plot, all trees with a DBH ≥ 10 cm were identified at species level and measured using compasses. The structural parameters of trees were calculated. The spatial data analysis tool was used to extract spectral indices from remote sensing images. In addition, road density, distance to road network and distance to water stream were calculated. Aboveground biomass was obtained using the allometric model. The conversion factor (CF = 0.50) was used to obtain the carbon stock. The Pearson correlation test was used to select the best environmental variable for carbon distribution modelling. The Radom Forest algorithm was used to predict the spatial distribution of aboveground carbon. The performance of model prediction was evaluated by cross-validation with 100 iterations. The coefficient of determination (R²), root mean square error (RMSE), prediction bias and Akaike Information Criterion (AIC) were used to measure model performance. Results: Results showed that the spatial distribution of carbon stock was strongly influenced by basal area, mean diameter, tree density, forest strats and water stress index. The relative importance of these variables was ranged from 4 to 96%. The model was very accurate (bias = -0.03%, RMSE = 3.24 t/ha and R2 = 98.93%). The quantity of carbon observed was ranged from 0.26 to 76.52 t/ha. While, the carbon predicted was varied from 0.24 to 64.66 t/ha for all stands. The carbon stock of forest stands was 2.11 times higher than that of the forest surrounding lands. The sites with high carbon stock were located in savannah woodlands. Anthropogenic pressure triggered the carbon storage potential of the FCGTR. Therefore, we had to strengthen the monitoring of plant species for sustainable forest management. |
