Statistical analyses
We first combined EAB presence data obtained from historical
publications with the two sets of field collected data (the field
surveys from 2003-2019, and the common garden field study from
2012-2013), then used a multinomial logistic regression (MLR) model to
analyze the relationships between degrees of EAB infestation (as ordinal
dependent variable) and host tree species, host tree age, DBH, habitat,
latitude, and altitude (as independent variables) collected from
different regions using the software JMP Pro 10.0.0 (SAS Institute
2012). Four variables, including latitude, altitude, DBH (continuous),
and host tree age (discrete) were binned to create categories using data
conversions presented in Table S3, respectively, to maintain consistency
of data processing. Host tree species, habitat and bark splits were set
as nominal variables, while host tree age, DBH, latitude, altitude,
canopy index and degree of infestation were set as ordinal variables.
One-way analysis of variance (ANOVA), followed by Tukey’s means
separation test, was used to evaluate effects of ash tree species and
habitat (as independent variables) on level of EAB infestation (as
ordinal dependent variable) in different regions using SPSS 21.0 (SPSS
Institute 2012). In the Oleaceae common garden experiment, ANOVA
followed by Tukey’s means separation test was used to evaluate effects
of ash tree species (as independent variable) on canopy index, bark
splits and percentage of infested trees (as dependent variable). When
determining the trend of EAB infestation level in relation to host tree
age, host tree size, and host tree geographic distribution mentioned
above, number of sample plots by age, DBH, and latitudinal and
altitudinal distribution of ash species were plotted by EAB infestation
level using OriginPro 2018 (OriginLab 2018).