This paper introduces a method for examining the scales of segregation
or aggregation of two kinds of plants. It uses the mapped positions o
f plants but is based on methods originally designed for contiguous qu
adrat data and may be referred to as two-dimensional local quadrat cov
ariance. A complementary randomization procedure makes it possible to
test whether observed aggregation is merely the result of overall patc
hiness. These methods are applied to analyze the spatial pattern of So
lidago canadensis L. plants in an old hayfield, with and without insec
t galls. In many of the 12 plots sampled, the galled and nongalled pla
nts were found to be aggregated at scales around 0.7m. The randomizati
on test showed, however, that this apparent aggregation was often due
to the overall patchiness of the plants' dispersion, and that in some
cases, within the patchiness the two kinds of plants were actually seg
regated at a similar scare.