Statistical Analysis
Statistics were analyzed with R statistical software (Development Core
Team 2013) and significance was attributed at P < 0.05.
The greenhouse frogs and pine woods treefrogs were analyzed together
because they were conducted simultaneously. Behavioral trials were
conducted separately for the Cuban treefrogs and southern toad and thus
they were analyzed using independent statistical analyses. For the Cuban
treefrog and the southern toad, we conducted general linear mixed
effects models (package: nlme, function: lme) with individual amphibian
as a random effect (to account for the lack of independence of repeated
behavioral observations on an individual), proportion of observations
avoiding during a trial as the response variable (arcsine square
root-transformed, unfortunately we could not get models with a binomial
error distribution to consistently converge), mass as a covariate, and
treatment and experience status (naïve versus experienced to Bd, nested
within individual as random effects) as crossed independent variables.
We followed these analyses with targeted hypothesis testing where we
pooled all treatments containing i ) any zoospores, ii )
dead zoospores, and iii ) metabolites in an effort to assess
whether any avoidance response was induced by the actual process of
infection, physical contact with Bd zoospores, or Bd metabolites,
respectively. There were too few greenhouse frogs and pine woods
treefrogs to have frogs assigned to treatments with controls solutions
assigned to both sides of the container as we did for the Cuban
treefrogs and southern toads. Thus, for the greenhouse frogs and pine
woods treefrogs, we tested for deviations between the observed and
expected null spatial distributions and tested for differences between
these two species in their responses because they were tested
simultaneously. Probability values were determined using the likelihood
ratio tests based on Type II sums of squares (package: car, function:
Anova). Amphibians that died during the experiment were excluded from
analyses.