Abstract
The identification of forest areas that are the sufficient sources of leaf-litter to streams is an important issue in riparian management, because leaf-litter is one of the most important foods for stream invertebrates. We monitored the leaf dispersal of two isolated chestnut trees in a deciduous forest on different sides of the stream, and then we verified the applicability of a simple physical model (modified Greene & Johnson (1989)’s model) to this actual data to predict leaf dispersal. The chestnut leaves were dispersed to about 25 m from the source trees, but most of the leaves fell within 15 m. The model predictions were consistent well with the observed results for the one tree, except for the vicinity of the source tree. Although the cumulative curves of the predicted leaf density were slightly lower than those of the observed leaf density, both the cumulative densities reached about 90% at 15 m from the source tree. We considerded that the model can estimate leaf dispersal with enough accurately when the wind regime is measured properly. Heterogeneous wind distribution in a riparian stand may impede accurate estimation of dispersal, but wind observation at a typical place in the stand is enough to get a rough estimation of leaf dispersal. Underestimation of leaf density near the source tree is not important problem for the purpose of this study. We are sure that this model is useful for estimating source distances of leaf-litter to stream in various types of riparian forests, although the model verification is insufficient.
