Ecological Interactions
After quantifying the environmental effects of the storm, I next sought
to characterize the consequences of those environmental differences for
survival and species interactions. Because attack byMordellistena and Eurytoma occurs in summer, they should
be unaffected by galls having been knocked to the ground. Conversely,
bird attack is likely to decline if galls are knocked to the ground and
are therefore less high off the ground (Start 2018a), and also tends to
decline further away from forest edges where woodpeckers live (Start et
al. 2018). I tested for these effects in two ways. First, I used an LMM
to estimate total mortality and attack rates of each enemy
independently, including year as a factor and population as a random
effect. I then performed pairwise comparisons of attack rates among
years using Tukey’s tests. Second, I used another LMM to estimate the
same measures of mortality and attack rates, but while including
distance to trees and percentage of stems knocked down as main effects.
In combination, these two tests (i) estimate among year differences in
demography and species interactions, and (ii) tie any differences to
specific environmental conditions.
While the above tests quantify mean differences among years and
environmental conditions, they do not estimate among-population
variation (i.e. whether the extreme event homogenizes the environment or
creates environmental variation). I again tackled this issue using
Levene’s tests. I independently tested for differences in
among-population but within-year variation in mortality and each attack
rate. I again compared years using Tukey’s tests to gain pairwise
comparisons. I also tied differences in variable species interactions to
environmental variation. Specifically, I calculated residuals for each
environmental variable and each ecological interaction. Residuals were
measured from the grand mean across all populations, but within a given
year. I then used linear models to estimate the residuals of each
species interaction using residual variance in the distance from trees
and the residual variance in the percentage of stems knocked down. As
above, this combination of tests quantifies among-population variance in
species interactions while tying that variation to corresponding
environmental variation.