To investigate applicability of the equilibrium landscape concept to various attributes of vegetation, the effects of sampling area (or grain size) on structural and compositional stand parameters were determined in an old-growth hemlock-hardwood forest. Three 1-ha plots, each gridded into I00 0.01-ha subplots, were established on the Roaring Fork watershed in Great Smoky Mountains National Park, Tennessee, USA. Estimates for 10 different stand descriptors were calculated for a variety of grain sizes (subplot aggregation levels) ranging from 0.01 to 1 ha. The stand descriptors included measures of physical structure (basal area and biomass) as well as measures of species composition (relative basal area).
All stand descriptors exhibited high deviation from the corresponding 3-ha mean at grain sizes approximating observed canopy gap area (< 0.02 ha). Deviations for total tree density, basal area and biomass diminished sharply with increasing grain size, while deviations for relative basal area of four important species remained relatively high even at grain sizes > 0.5 ha.
The relationship between sample variance and grain size was inverse and approximately log-log linear for all descriptors. Biomass, basal area, and large canopy tree density had relatively steep slopes. These variables of physical structure presumably were related to canopy gap size and distribution. The remaining measures of density and relative basal area had gentler slopes, indicating a milder decrease in variation with increasing grain size. Coefficients of variation for each parameter also showed this response to grain size, with compositional parameters having relatively high variation at scales > 0.5 ha.
In general, the nature of physical structure patches (e.g. total basal area and biomass) differed from that of composition patches (e.g. relative basal area of individual species). This contrast should be considered in equilibrium landscape concepts and vegetation sampling design.