In this study, researchers quantified the density and complexity of understorey flora of oldgrowth forest and of selectively logged, fragmented and regrown forest. They determined various parameters from standardized photographs from these forests, and calculated correlation between these parameters.

The researchers selected two sites from a forest in Malaysia (primary and logged) and randomly selected 14 study points from grids of each of the sites. They then took photographs in the N, S, E, and W directions from the points. The photographs were then displayed such that only the borders and interiors of vegetation became significant.

The parameters were:

• core area: (total are of vegetation patches) - (outermost pixel from all patch edges)
• number of patches: number of discrete vegetation patches
• mean patch size: mear area of vegetation patches
• patch size standard deviation
• total edge: sum of the edge lenghts of vegetation patches
• landscape shape index: (sum of edge segments) / (square root of total area of image)
• average weighted mean shape indes: (patch perimeters) / (square root of patch areas) and
• average weighted mean patch fractal dimension: 2*log(patch perimeter) / log(patch area) * (patch area) / (sum of the area of all patches)

The parameters measuring patch density autocorrelated strongly, as did those for patch complexity. However there were no significant correlations between the parameters for density and complexity.

In general, the logged forest tended to be denser than the primary forest, partly because the logged forest resembles areas of primary forest in which natural treefalls cause a naturally denser understorey. The logged forest showed much less variance in density and complexity, and thus indicated a loss of heterogeneity.

Analyzing understorey density and comlexity in this fashion could be an important tool in studying animal and plant habitat associations.