Background and aims Several investigations have assessed the soil organic carbon (SOC) content in forest soil. However, very few studies have explored the spatial variability of SOC content in forest soil using deterministic and geostatistical techniques. Telangana hosts various forest types, including dry deciduous scrub, dry teak forest, southern dry mixed deciduous forest, and non-forest areas. The present investigation aims to identify the best-fit model for SOC content distribution and examine how various forest types influence SOC content in forest soils in Telangana, India.
Methods The air-dried soil samples were analyzed for their SOC content using the wet digestion method. Deterministic and geostatistical methods used to assess the spatial distribution of SOC content in unsampled regions. Data were utilized to create spatial SOC maps using five interpolation methods: Inverse Distance Weighting (IDW), Ordinary Kriging (spherical, gaussian, and exponential), and Empirical Bayesian Kriging (EBK). The accuracy of these models was evaluated through cross-validation, semivariogram and considering metrics like coefficient of determination (R2) and the mean error (ME) and root mean square error (RMSE).
Results The order of SOC content was observed as follows: southern dry mixed deciduous forest > dry teak forest > dry deciduous scrub > non-forest for all soil depths, except at 30–60 cm. The results indicate that the EBK model has the highest R2 value (0.228) followed by OK- Spherical (0.219) and Exponential (0.216) for a soil depth of 0-30 cm. The OK-spherical model has the highest R2 value (0.139) followed by OK- Gaussian (0.135) and EBK (0.132) for a soil depth of 30-60 cm, and IDW has the highest R2 value (0.168) followed by OK- exponential (0.144) and EBK (0.135) for a soil depth of 60-90 cm.
Conclusion The study examined soil organic carbon (SOC) content and its spatial distribution across various forest types in Telangana, utilizing deterministic and geostatistical methods. Southern dry mixed deciduous forests exhibited higher SOC content, whereas lower SOC content were observed in dry deciduous scrub and non-forest areas. Evaluation of interpolation methods indicated that geostatistical methods (EBK and OK) outperformed IDW at soil depths of 0-30 cm and 30-60 cm, while the deterministic method (IDW) performed well at a depth of 60-90 cm. Increasing sampling points and incorporating elevation or topographical information could improve the accuracy of the interpolation model, emphasizing the importance of using both deterministic (IDW) and geostatistical (OK and EBK) methods for generating the SOC content spatial distribution map in the entire study area.