Abstract:

Aims: Tree mortality is a natural demographic process that plays a key role in determining forest dynamics and succession. Elucidating spatial patterns and driving factors of tree mortality would help researchers to better predict forest dynamics. In this respect, studies of the relationships between dead individuals and their habitats can aid the analysis of spatial distribution patterns. Previous studies on tree mortality have focused mainly on the causes and processes of tree death, temporal and spatial variabilities, impacting on ecosystems, and the abundance patterns and death dynamics of snags. Therefore, this study aims to determine the distribution patterns of dead individuals at different scales and the relationship between snags and habitats in a karst forest.

Methods: Using the data collected from two surveys of a 25 ha forest dynamic plot in the Mulun National Nature Reserve in Guangxi, China, we analyzed the species composition, size class structure and spatial patterns of snags, and its association with habitats. First, the univariate pair correlation was employed to analyze the distribution pattern of dead individuals of different species. Then, the Torus-translation method was utilized to analyze tree mortality and habitat-type associations.

Results: A total of 17,306 snag individuals, belonging to 57 families, 130 genera, and 194 species were found in the plot. The three families with the highest numbers of dead trees were Alangiaceae, Rosaceae, and Flacourtiaceae, with Alchornea, Pyracantha, and Alangium being the top three genera and Alchornea trewioides, Mallotus barbatus, and Mallotus japonicus the top three species with the highest mortality rates. The average diameter at breast height (DBH) of the snags was 3.83 cm, with the maximum DBH being 47.11 cm. The snags had an inverse J-type tree size structure. Within the scale of 0-50 m, the snags were mostly aggregated in the plot, but some large-sized trees were randomly distributed. The numbers of species whose mortality were positively associated with a hilltop, steep slope, gentle slope, and depression were 41, 13, 41, and 38, respectively, whereas the numbers of species with a negative association with the same habitat types were 38, 67, 33, and 10, respectively. Those of species with no association with hilltop, steep slope, gentle slope, and depression were 4, 3, 9, and 35, respectively. Among the 20 most dominant trees in the area, all were found to have either a positive or negative association with the four habitat types, with the mortality of each size class being positively correlated with hilltops but negatively correlated with gentle slopes.

Conclusion: Our study revealed that the distribution of snags in the surveyed plot was aggregated in a karst evergreen and deciduous broad-leaved forest in the Mulun National Nature Reserve,. This indicates that tree mortality was nonrandom and more highly impacted by the habitat type and topography.

Key words: tree mortality, DBH class structure, point pattern analysis, habitat association, karst ecosystem