The Yellow Meadow Ant Lasius flavus: spatial analysis of its nest distribution
Statistical Method Two: Nearest Neighbour Analysis
Choosing an appropriate quadrat size for this technique does cause problems, in that density
calculations are a necessary pre-requisite here. Equally important, however, is that accurate
measurements from one point to its next nearest neighbour arc taken. This measurement was taken
from the centre summit of each anthill. This is best carried out with the individual anthills being
plotted onto graph paper using a co-ordinate system in the field.
1) Firstly all the anthills Were assigned a number and their nearest neighbour was identified
together with the straight-line distance between them being noted.
2) Next all the distances were summed and the result divided by the number of points (74
for Yardley Hill and 21 for Daisey Plain) in order to calculate the mean nearest neighbour
distance. As students worked in pairs and started from different reference points within
the 20m x 20m sample area, this was replicated 6 times and a mean result was derived of
1.3107m and 2.7m for Yardley Hill (south-facing) and Daisey Plain (north-facing)
respectively (see Table 2).
3) In order to see whether the result obtained is what we would expect if the anthills were
scattered randomly (based upon no interaction) we need to know the size of the area from
which the data were collected (in this case 20m x 20m = 400 square metres). Next this
value is used to calculate an expected value for a random distribution (Nnran) for each
site.
For Yardley Hill (south-facing):
Nnran = 1/(2 x VP) where P = the density of points (no. per unit area*)
(P- 74/400 - 0.185)
Nnran = 1/(2 xVo. 185)
= 1/2x0.4301 = 1/0.8602
= 1.162 metres
For Daisey Plain (north-facing):
Nnran- 1/(2 x VP) where P = the density of points (no. per unit area*)
(P =21/400 = 0.0525)
Nnran- 1/(2 x VO.0525)
- 4.3644 metres
(*Area of sample area = 20m x 20m = 400 m2)
4) The next step is to calculate the Nearest Neighbour Index (NM). The index gives a
single measure of the distribution of points (anthills) on a scale ranging from 0 (clustered)
through to 1 (random), to 2.15 (dispersed). It is found by dividing the observed value
(Xnn) by the random distribution value (Nnran):
Essex Naturalist (New Series) 19 (2002)
111