3.2 Plant trait trends across soil type, species, and
climate
Analyses using randomForest were performed using species and
community level traits. For the species-level, analyses were performed
with soil type, species identity, and climate as response variables.
Species was most strongly associated with maximum height, leaf area, and
leaf dry weight (22.3 %, 17.8 %, and 16.47 % respectively), with a
38.76 % error decrease. Soil was most strongly associated with
mortality, dry weight, and number reproducing (8.5 %, 4.5 %, and 4.0
% respectively), with a 35.93 % error decrease. Climate was most
strongly associated with biomass, maximum height, and leaf area (16.29
%, 12.92 %, and 12.1 % respectively), with a 38.33 % error decrease.
Overall, species’ trait responses were more consistent across climate
than soil, with the highest variable predictive ability being observed
in association with species. For the community level, analyses were
performed with soil and climate as response variables. Soil type was
most strongly associated with mortality, root biomass, and reproduction
(24.56 %, 20.00 %, and 19.72 % respectively), with a 26.67 % error
decrease. Climate was most strongly associated with leaf area, leaf dry
weight, and mortality (39.78%, 28.57%, and 27.68% respectively) with
a 46.63 % error decrease. Once again, climate was observed to have more
predictable trait responses than soil. For the site level, analyses were
performed with soil type and climate as response variables. Soil type
was most strongly determined by mortality, reproduction, and maximum
height (39.42 %, 39.17 %, and 18.75 % respectively), with a 41.63 %
error decrease.