Spatial pattern of vegetation in a subtropical savanna landscape. Liu, Feng^*,1, Wu, X.Ben¹, Bai, Edith¹, Archer, Steve², Boutton, Thomas¹, Jessup, Kirk¹, ¹ Department of Rangeland Ecology and Management, College Station, TX² School of Natural Resources, Tucson, AZ

ABSTRACT- Scaling addresses how ecological patterns and processes operate at different spatial and temporal scales, and is fundamental to ecological extrapolation and assessment of land management practices. Detecting the domain of scales is an important first step in elucidating scale-dependent properties of ecosystems. The purpose of this study is to characterize the domain of scales of vegetation and to quantify spatial variation in the distribution of woody species along a topographic gradient in a subtropical savanna landscape in southern Texas. Upland portions of this landscape are comprised of discrete woody patches embedded in a grassland matrix, while lowlands are comprised of closed-canopy woodlands and closed-basin playas. A 309 m upland-to-lowland transect was established along this catena gradient. Herbaceous and litter biomass were collected in 0.5 m x 0.5 m plots at 1 m intervals along the transect. Presence and basal diameter of trees and shrubs were mapped in 2 m (shrubs) and 12 m (trees) belts centered on the main transect. Several statistical methods were used to quantify the spatial pattern of vegetation along the transect. Quadrat variance methods (local quadrat variance, new local variance, and paired quadrat variance) indicated a 45 m scale domain in grass, forb, and litter biomass and shrub density. Fractal and lacunarity analyses confirmed those results. Woody species differed in their spatial distribution along the topographic gradient. This knowledge of scale domain will facilitate our efforts to extrapolate ecosystem patterns and processes to landscape and regional levels of organization.

KEY WORDS: Spatial variation, Scaling, Domain of Scales, Quadrat variance method