Statistical Analyses
To test the hypothesis that burned habitat was of nutritionally higher quality than unburned habitats, we analyzed the nutritional quality data in a hierarchical fashion. For the analysis, we only used species that were found in both habitat treatment types. Four species that were only found in the burned habitat (fireweed, quaking aspen (Populus tremuloides ), little tree willow (Salix arbusculoides ), and sedges) were removed from the analysis, leaving nine different species to be included in the analysis (Alnus crispa , Equisetumspp., Salix pulchra , S. glauca , S. pseudomyrisinites , S. richardsonii , Betula nana , B. glandulosa , and B. neoalaskana ). We first compared overall nutritional quality of habitats for all forages combined using a multivariate analysis of variance (MANOVA) with habitat type, month, and year as covariates. We tested the interaction between habitat type and month sampled using an ANOVA, with digestible energy, nitrogen concentration, percent reduction in proteins from tannins, and digestible protein as dependent variables. We also calculated the species diversity of each habitat type using the Shannon-Weiner Index (Eq. 2), where pi is the proportion of total sample represented by species i, and H is the diversity of each study site.
H=Σ[(pi)×ln(pi)]. (2)
Comparisons of forest versus burn for individual forage species common to both sites were done using unpaired t-tests with a Bonferroni multiple comparison adjustment. All statistical analyses were performed in R (Marsh et al. 2006), and all comparisons were considered significantly different at p < 0.05. Individual species that were found in both burned and unburned habitats, including diamond leaf willow, dwarf birch, and gray leaf willow, were also compared individually using unpaired t-tests with a Bonferroni multiple comparison adjustment to directly compare the two habitat treatment types.
Results
Over the 2018-2019 field seasons, we sampled the Alphabet Hills burn on six occasions, five times across the summers of 2018 and 2019 and once in the winter of 2018-2019. We collected over 500 plant samples, and measured canopy coverage of moose browses from randomly selected plots in burned (n = 16) and unburned habitats (n = 10). We found that the total canopy cover of moose preferred browses were higher in forested sampling sites than in burned sites. There was a significant difference between browse canopy cover in the forest compared to the burn (forest % cover = 10%, SE = 0.019, burn % cover = 9%, SE = 0.01; t = 19.18, p = 0.033) (Figure 2). The burned area had higher diversity of species (burn H = 1.709, forest H = 1.509) with four species that were only found in the burned areas including fireweed, quaking aspen, little tree willow and various mushrooms (Boletus spp.) that are consumed by moose (Figure 2).