2.1 | Plant species and deer browse
The four species included in this study were the indigenous treeNyssa sylvatica Marshall, the indigenous shrub Lindera benzoin L. Blume, the nonindigenous, invasive shrub Euonymus alatus (Thunb.) Siebold, and the nonindigenous, invasive semi-woody shrub Rosa multiflora Thunb. We selected them based on four criteria. First, they were sufficiently abundant in the understory to provide a sample of individuals in both fenced and unfenced plots. Second, they included a mix of indigenous and nonindigenous, invasive species, since both types are common in suburban forests and comparisons of their ecologies is relevant to a broader understanding of plant invasions. Both invasive species are of conservation concern . Third, they include both a tree species and shrubs. The ecological success of both groups is essential for maintaining the physical layers of forest structure and food sources that support a diversity of other forest species . Fourth, the four species were all vulnerable to deer browse, but at somewhat different rates.
Deer browse rates for each species were measured on unfenced and fenced plants in the summer of 2018, the season when we sampled the leaves for metabolomic analysis, and over multiple seasons from 2012-2019, including one fall, two winters, and seven summers. The rates were calculated in each plot by inspecting all individuals of the species in an L-shaped, 0.5 x 7.5 m belt transect that followed two edges of the plot, and recording the presence or absence of the distinctive, tell-tale signs of deer browse: bitten, shredded twig tips . Any browse signs inside the fences should be considered as the error rate of falsely assigning a damaged twig tip to deer browse when it was due to some other cause. For the browse rates presented for the species in this study, the fenced rate on the species was subtracted from its unfenced rate. The browse rates were compared among species for the summer 2018 rates and the rates pooled across years, using G-tests for overall heterogeneity among the species along with pairwise tests with a Bonferroni correction for multiple tests on the same data set. G-tests were done in R v 4.1.2 , with the G.test function in the RVAideMemoire package.