Species-habitat association analysis is useful to detect spatial arrangement of individual plants, to discover rules about the distribution of species and to generate hypotheses about the possible underlying process controlling observed structures. Quantifying methods were used to classify habitats in terms of topographical variables in a mixed temperate broad-leaved Korean pine forest of the Changbai mountains in northeastern China. All of the 625 20 m × 20 m quadrats of the plot could be unambiguously assigned to one of three habitat categories (low-plateau, high-plateau and slope). Torus-translation tests were used to estimate species-habitat associations. Many species are clearly distributed in a biased fashion with respect to habitats. Fifteen (55.6%) out of 27 species showed strong positive or negative association with specific habitats. We compared species-habitat associations at the sapling and adult stages. Adjusted density values indicated few species exhibit extremely strong habitat associations. Only 9 out of 26 species had adjusted densities > 3 in the habitat for which they had strong positive affinity. Few species show the same associations at the small tree and large tree stages. Only 3 out of 22 occurring associations with a specific habitat appeared to have a consistent habitat association at the two stages. These results suggest that species-habitat associations exist in the 25-ha plot of the temperate forest of the Chang- bai mountains. Owing to limitations in our statistical methodology, we partly underestimated associations by ignoring rare species. Regeneration niches can contribute to co-existence, but regeneration niches due to habitat associations play a limited role in species co-existence, since most species show a similar trend in habitat associations at the sapling and adult stages. We should pay more attention to shifts in habitat associations, i.e. niche shifts at different stages of existence.
