Abstract: The study shows that the snags of L.gmelini are resulted from fire, suppression, or ageing. There are more medium diameter snags in L.gmelini Betula fruticosa and L.gmelini Ledum palustre forests, more medium and large diameter snags in L.gmelini Herbae forest, and more medium and small diameter snags in L.gmelini Alnus mandshurica forest. For dead trees, the uprooted trees are dominant in L.gmelini Betula fruticosa and L.gmelini Alnus mandshurica forest, snapped trees are dominant in L.gmelini Ledum palustre and L.gmelini Herbae forests. The treefall direction is chiefly affected by topography, run of mountain, and direction of dominant wind, which is random in L.gmelini Betula fruticosa forest, east in L.gmelini Alnus mandshurica forest, south and northeast for uprooted trees in L.gmelini Ledum palustre and L.gmelini Herbae forests, respectively, and random for snapped trees in both forests. In L.gmelini Herbae and L.gmelini Alnus mandshurica forests, the density of seedlings and saplings on fallen dead wood (18 and 40 individuals per m², respectively) is greater than that on mineral soil (2 individuals per m²).The regeneration of seedlings and saplings is quite good (28 individuals per m²) on dead logs and on forest floor in L.gmelini Betula fruticosa forest. The effect of dead logs on regeneration is not clear in L.gmelini Ledum palustre forest.

Key words: Larix gmelini, Old growth forest, Snag and dead log, Regeneration, shrub ecosystem, rocky desertification, plant nutrient, soil nutrient.