The third canonical variate, likewise independent of
the first 2 and of less importance, consisted of a positive
association between the number of Agasicles and internode
length and the number of stems. We can conclude from this
that adult Agasicles tend to occur more frequently in dense
stands of alligatorweed with samller stems, and the latter
condition is negatively correlated with the number of Vogtia
and with Agasicles feeding. The Agasicles feeding and stem
size association is in agreement with the first canonical
variable but much smaller in magnitude. The positive
correlation of Vogtia with stem size is expected since in
small stems the full grown larvae may be as large in diameter
as the stem itself and, therefore, not protected from para-
sites and predators or environmental factors.
It should be noted that 63.96% of the variation is
explained by canonical.variable 1,. 30.92%. by canonical:
variable 2, and only 5.12% of the variation is explained
by canonical variable 3. The most important associations,
therefore, are probably described by canonical variables
1 and 2.
In the second analysis, the same 4 plant characteristics
i.e. plant height, internode length, internode diameter, and
number of stems were used to compare with the number of
Agasicles, Agasicles feeding damage and each of the 7 immature
stages of Vogtia present.
In this correlation, 4 new canonical variables were
formed (Table 8). The first 2 were significant at the .01