Statistical Evaluation of Litter Interception Models Across Forest Sites Using Regression, ANOVA, and Cluster Analysis

Abstract: Forest litter interception is a key hydrological process influencing water balance and ecological resilience. Litter Interception refers to precipitation that does not reach the soil, but instead intercepted by the leaves, branches of plants and the forest floor. This study evaluates precipitation–litter interception relationships across four forest sites located in Nongmaiching Hill using linear regression, ANOVA, and hierarchical clustering. The interception magnitude showed the order Broad Leaved Forest>Logged Forest>Pine Plantation>Secondary Growth Forest.Regression diagnostics revealed varying model strengths,Site II i.e. Pine plantation showed the strongest model fit (R² ≈ 31%, p = 0.0012) showing the highest explanatory power , followed by Secondary regrowth forest following shifting cultivation Site IV (R² ≈ 22%, p = 0.0073) demonstrating efficient predictive accuracy .Sites I and III Broadleaved Natural and Logged forest sites exhibited comparativelyweaker but significant fits. Heterogeneous litter structure and complex broadleaved canopymight be responsible for such results. Cluster analysis groupedtogether Broadleaved Natural and Logged forest sitesat the lowest linkage distance showingthemost similar sites merge at a low distance exhibiting the close connection indicating strong similarity in their interception dynamics whereas Pine plantation and Secondary regrowth forest Sites formed a distinct cluster reflecting differences in hydrological resilience. Cluster analysis validated regression-based site differences. The study suggests that structurally diverse or mature canopies may involve more complex interception processes and dynamics. Site-specific variability in litter interception provides a comparative framework for forest management and policy applications.