Comparison of distance based density estimates for some arid rangeland vegetation [Australia]

A method was required for determining the effect of management on extensive populations of trees and shrubs in central Australian rangelands. One useful indicator of change in these populations is the density of individuals, and there are several methods available based on distance measurement for density estimation. This study compared those procedures. Samples were drawn by computer from ground maps of actual plant distributions for Acacia aneura, Cassia nemophila and Atalaya hemiglauca and from a map generated at random. These samples were drawn to examine the properties of the nearest neighbour, point centred quarter, conditioned distance and compound T‐square estimates of density. Samples were drawn by two methods: simple random sampling and semisystematic sampling. In general, there was a tendency for all estimators of density to underestimate the true density of naturally occurring populations with the compound T‐square method (Byth 1982) being most robust. The compound T‐square method was least biased but its variance increased for more aggregated spatial distributions. Estimates of density were not altered by the use of semisystematic sampling, when compared to simple random sampling. The spatial distributions examined in this study have not previously been studied as theoretical models. Acacia aneura and Cassia nemophila showed some aggregation of clusters, while the Atalaya hemiglauca showed a more extreme form of clustering due to its root suckering propagation.