S. Mehlert et Gr. Mcpherson, EFFECTS OF BASAL AREA OR DENSITY AS SAMPLING METRICS ON OAK WOODLAND CLUSTER ANALYSES, Canadian journal of forest research, 26(1), 1996, pp. 38-44
This study compared the effects of sampling metrics (basal area and de
nsity of trees) on cluster analyses of oak woodland vegetation. Minimu
m-variance clustering was used to define communities within oak woodla
nds. Communities derived from basal area data were more distinct and m
ore homogeneous than communities derived from density data. Consistent
with previous research, minimum-variance clustering of basal area dat
a produced groups of approximately equal size. In contrast, minimum-va
riance clustering of density data did not result in similar-sized grou
ps. Formation of similar-sized groups, which is generally reported to
be an advantage of minimum-variance clustering, apparently is dependen
t on the sampling metric employed.