Fig 1.
Average annual reported cases of Lyme disease (LD) for 2012–2015 in the US Lake States.
Cases reported per 10,000 county residents.
Fig 2.
Average annual reported cases of deer-vehicle collisions (DVC) for 2012–2015 in the US Lake States.
Cases reported per 10,000 county residents.
Fig 3.
Location of P2-plus plots (n = 786, 2012–2015) within the US Lake States with corresponding deer density.
Deer density provided by QDMA. Plots located outside the delineated state borders occurred on islands of the Lake States.
Fig 4.
Location of P2-plus plots (n = 792) within the Lake States with browse impact score, 2012–2015.
Plots located outside the delineated state borders occurred on islands of the Lake States.
Table 1.
Browse impact assessment for FIA Phase 2-plus plots and number of FIA plots within each state measured between 2012–2015.
Table 2.
Summary statistics for forest inventory variables used in randomForest analysis from P2-plus and P2 plots.
Table 3.
Models for US Lake States region created using random forests analysis with explanatory variables listed in order of importance predicted by the models.
Table 4.
Models for each individual state created using random forests analysis with explanatory variables listed in order of importance predicted by the models.
Fig 5.
Model accuracy for models created by random forests for the US Lake States.
Proportion of plots with browse scores predicted correctly based on models created by random forests models for the Lake States, Minnesota (MN), Wisconsin (WI), and Michigan (MI). Corresponding models found in Tables 3 and 4. Error bars depict ± 1 SD from mean.
Fig 6.
Inverse distance weighted interpolation of deer browse pressure in US Lake States.
Browse score based on predicted browse score provided by the full Lake States model. Browse impact score aligns with values described in Table 1. Green indicates the first quantile of predicted browse score, yellow as the second quantile, and red as the third quantile with cutoffs at 0.33 and 0.67 quantiles. White areas indicate insufficient forest cover for prediction.