Understanding the dispersal process of leaf litter is important for the estimation of spatio-temporal variation in nutrient cycling.  In forest community with diverse species, leaf fragments disseminated to forest floors are often observed as litter mixtures with several species.  Because of difficulty in identifying the sources (mother trees) of leaf fragments, inter-specific variation in leaf dispersal and its association with tree functional traits remain poorly understood.  In this study, we developed a spatially-explicit integrative model that describes the production and dispersal kernel of leaves, by using data both of the spatial distribution pattern of tree stems(source) and the abundance and species composition of leaf litter mixtures (sink). We applied this model to dataset of major tree species in riparian forest community, and tested the associations of estimated leaf dispersal distance with functional traits of tree species.  The mean leaf dispersal distances were estimated longer for deciduous oak and beach in the tested site.  The association of leaf dispersal distance and traits of species was detected for tree height, while associations with leaf traits were rarely detected.  Our study is useful to unify the community structure of trait assembly and abiotic (environmental) process of nutrient cycling through decomposition.