Estimating density and dispersion from truncated or unrestricted joint point-distance nearest-neighbour distances.

Forty computer-simulated populations were analysed to derive formulae for estimating density of populations from a set of distances from sample points to the nearest member, from that member to its neighbour and from that neighbour to its nearest neighbour. Distances may be truncated or unlimited.
The formulae were applied to data from 11 paper-dot and 18 field populations in a total of 37 experiments. Fifty-nine percent of the corrected estimates were within 10 percent of "true density", 76 percent were within 20 percent, and 93 percent were within 30 percent. Most of the unacceptable estimates were attributable to sampling difficulties (particularly in paper dot populations) or sampling errors of the "true density" values.
An index of non-randomness is an intrinsic part of the density estimating formula. This index is described, and values for the experimental data are given to illustrate the spectrum which can be expected in biological populations