Main content area

Spatiotemporal differences in tree spatial patterns between alluvial hardwood and mountain fir–beech forests: do characteristic patterns exist?

Janik, David, Adam, Dušan, Hort, Libor, Král, Kamil, Šamonil, Pavel, Unar, Pavel, Vrška, Tomáš, Horal, David, Adler, Peter
Journal of vegetation science 2013 v.24 no.6 pp. 1141-1153
data collection, floodplains, hardwood, hardwood forests, montane forests, mortality, rivers, trees, Czech Republic
QUESTIONS: What are the differences between the tree spatial patterns (TSP) of various recruit and mortality waves in alluvial hardwood forests and mountain fir–beech forests? Are there any statistically significant differences between the mean TSP of these forest types? Are these differences stable over time? LOCATION: Alluvial floodplain forests at the confluence of the Morava and Dyje rivers, and mountain fir–beech forests in the Outer Western Carpathians, Czech Republic. METHODS: In both forest types, seven 2‐ha rectangular plots were analysed. The pair correlation function g(r) was used to describe tree density variability of trees with DBH ≥ 10 cm. The analyses were carried out for data sets from the 1970s, 1990s and 2000s. A bootstrap method was used to test for significant differences between the mean values of g(r) from alluvial forests and from fir–beech forests. RESULTS: Recruits in mountain fir–beech forests revealed consistent clustering to at least 5 m. In alluvial hardwood forests, recruits also showed random distribution as well as occasional regular distribution at distances over 20 m. Bootstrap significance tests revealed significant differences between the mean values of g(r) for alluvial forests and fir–beech forests. Alluvial floodplain forests showed statistically significant stronger clustering up to a distance of 4 m in all study periods. At distances over 20 m, mountain fir–beech forests demonstrated stronger clustering. In the 1970s, this was statistically significant only at a distance of 32 m, but in the 2000s, it was at intervals of 22–30 and 34–38 m. CONCLUSIONS: The methods of data analysis in this study enabled us to find significant features of TSP at finer resolution than the common resulting trichotomy of univariate analysis: clustering, randomness and regularity. We believe that, on the basis of detailed spatial analyses, it is possible to create a TSP model that reflects the typical features of particular forest types.