Variables influencing humus forms differentiation in Hyrcanian forest in the local scale (Case study: Shast-Kalateh Gorgan)

This study aims to identify factors affecting the secondary humus forms differentiation in the East Hyrcanian forests. For this purpose, we select 189 sample points using a nested transect sampling design in the Dr-Bahramnia forest. Humus forms identified based on morpho-functional classification and adjacent trees characteristics were recorded and then forest type were determined. Primary and secondary topographic attributes and land forms were extracted from digital elevation model due to the geographical location of sample points. Random forest classification, boosted tree classification and multiple adaptive regression Spline classification were applied as modelling tools to analyse important variables influencing differentiation humus forms. The seven secondary humus forms were differenced include Dysmull, Eumull, Mezomull, Hemimoder, Dysmoder, Eumoder and Eumacroamphi. The highest accuracy and agreement between the observed and predicted were boosted tree classifier while adjusted rand index and Cohen’s kappa coefficient were 0.76 and 96.3 percent respectively. The results showed that humus forms variability were dependent on the characteristics of the surrounding trees of sample points and topography.