Soil quality dynamics in natural pine forests lands integrating fuzzy-AHP and prediction based on random forest...

Soil quality assessment is crucial for monitoring and restoring soil functions and ensuring the soil's ability to support sustainable production. The present study aimed to assess the soil quality of a pine-forested region in the Engiz basin of Samsun province, Türkiye. We applied the minimum dataset (MDS) and total dataset (TDS) indicator selection method and linear and non-linear scoring approach and integrated with the Fuzzy-analytical hierarchical approach (F-AHP) to evaluate the soil quality of the region. Principal component analysis (PCA) reduced the initial set of 28 soil quality indicators to 12 most representative indicators, namely, sand, silt, structural stability index, organic matter, calcium, potassium, calcium carbonate, copper, manganese, soil respiration, and carbon-to-nitrogen ratio. Regardless of scoring techniques, soil quality obtained based on MDS adequately represented the TDS approach, with a significant correlation coefficient (r > 0.85, P < 0.01) and strong linear association (R² > 0.64). Non-linear (NL) models consistently performed better than linear models, and TDS_NL (sensitivity index (SI): 2.29) emerged as the best model in representing the soil quality of the study area, followed by MDS_NL (SI: 2.23). Repeated 10-fold cross-validation (with 3 random repeats) results showed that random forest models accurately predicted soil quality across all soil types (R² > 0.75), emphasizing their utility in soil quality evaluation studies. Both the observed and predicted soil quality maps, regardless of the indicator selection or scoring method, showed a consistent spatial trend, with lower soil quality mainly concentrated in the southern and southwestern areas, moderate soil quality in the central area, and higher soil quality in the north and northeastern regions. The results of our study approach are expected to offer valuable insights into sustainable forest soil and forest use management in forest-dominated landscapes in similar ecological and climatic conditions.