Spatial pattern of woody vegetation in a temperate deciduous forest

Jeffrey Stuart Ward, Purdue University

Abstract

The primary objective of this research was an examination of the dynamics of tree spatial distribution patterns. Ten indices were evaluated to determine their relative utility in describing spatial pattern: Clark-Evans' plant-to-plant index-R$\sb{\rm CE}$, Pielou's point-to-plant index-a$\sb{\rm p}$, Holgate's 1:2 ratio-H$\sb{12}$, Holgate's 3:4 ratio-H$\sb{34}$, Chapham's variance:mean ratio-V:M, Grieg-Smith's hierarchical ANOVA-GS, Hill's two-term local variance-TTLV, Morisita's index of dispersion-I$\sb{\rm M}$, the Non-parametric dispersion index-IDPN, and Annuli density analysis-ADA. Indices' biases were evaluated using simulated random dispersions at eight density levels: 20, 40, 60, 80, 100, 200, 300, and 400 stems/hectare. Indices' means and/or variances were found to be biased by population density. Distance indices were biased by edge effects and most area indices were biased by a scaling factor. Three uniform and nine or fifteen aggregated dispersion models were combined with five population densities to evaluate the power (1-Type II error) of each index in detecting non-random spatial patterns. R$\sb{\rm CE}$ and a$\sb{\rm p}$ were powerful for both aggregated and uniform spatial patterns. H$\sb{12}$ and H$\sb{34}$ were not powerful for uniform patterns. V:M. GS, I$\sb{\rm M}$, and IDPN were powerful in detecting non-random spatial patterns. IDPN was the best method for describing spatial pattern scale. V:M, GS, TTLV, and I$\sb{\rm M}$ were not able to distinguish between aggregations of 28 m$\sp2$ and 1810 m$\sp2$. ADA had high power in describing pattern scale when it determined a pattern was non-random. In 1986 all stems ($\geq$10 cm dbh) were mapped in a 4 ha section of the Davis-Purdue Research Forest. Stems in this area had been mapped in 1926 and 1976. Stems 0-10 cm dbh were inventoried on a 10 $\times$ 10 m grid. Canopy structure was classified on a 5 $\times$ 5 m grid. There was a significant trend towards the establishment and maintenance of a uniform spatial pattern for stems $\geq$10 cm dbh at the community level. Many species were aggregated at scales, $>$1000 m$\sp2$, which are probably related to soil drainage. Species' regeneration was found to have distinct responses to canopy structure which closely corresponded to published understory tolerances. The past low-intensity disturbance regime has created a canopy structure with a high percentage of primary canopy to canopy gaps. This favors development of tolerant understory species which characterize late-seral communities. (Abstract shortened with permission of author.)

Degree

Ph.D.

Advisors

Parker, Purdue University.

Subject Area

Forestry

COinS