Figure 2. Both microhabitat parameters that are not expected to
be directly affected by climate change (a-d) or expected to directly
respond to climate change (e-i) differed in their within-site
variability and show variable relationships with elevation. We show
within (σ2within) and among
(σ2among) site percent variation
explained. These variance partitioning results are from Linear Mixed
Models (LMMs) of the microhabitat parameter on the y-axis and block
nested within site (microhabitat parameter ~ 1 +
(1|site/block). In cases of model fitting issues, we ran the
LMM with only (1|site) as a random effect and thus only report
among-site percent variation explained. We also fit LMMs to test the
effects of quadratic elevation, transect, and their interaction on the
microhabitat parameter indicated on the y-axis, with a random effect of
site (microhabitat parameter ~ poly(elevation, 2) *
transect + (1|site)). We show one fitted line for models for
significant (alpha = 0.05) effects of elevation (line is shown for MB
transect only), and two fitted lines for significant elevation*transect
interaction even if elevation itself was not significant. Note that
fitted lines do not include random effects. Legend for all plots is as
in (a) with different colors indicating the warmer, wetter West (MB)
transect or the cooler, drier East (RP) transect.