Statistical analysis of the distribution pattern of natural regeneration in forests

Abstract

This paper describes the analysis of spatial patterns formed by natural regeneration in mixed mountain forest stands. The coordinates of the plants were taken in circular quadrats which were regularly arranged in plots distributed within the research areas. We analysed first univariate patterns, both for all regenerations pooled together and for the main tree species separately, and second bivariate patterns for all combinations of main tree species. The patterns were described by indices measuring the deviation from randomness at different scales—from nearest-neighbor to whole-quadrat level. For each index we performed a three-step test procedure which gave a multiple error probability for the null hypothesis of overall randomness. The analyses showed no pronounced deviation from a random distribution within quadrats of 1 m 2, although the univariate patterns tend to a very slight aggregation. At larger scales we observe significant univariate aggregation and bivariate segregation. Bivariate association may also be present, but could not be confirmed.