Do wolves protect forests? Investigating the link between wolf density, deer browse, and plant recovery
Data files
Oct 14, 2025 version files 454.24 KB
-
Analysis_Code.Rmd
32.34 KB
-
Browse_Data_Goldenrod.csv
36.22 KB
-
Browse_Data_Oaks.csv
191.61 KB
-
Covariates.csv
3.57 KB
-
Growth_Data_Goldenrod.csv
49.72 KB
-
Growth_Data_Oak.csv
89.80 KB
-
PlantCommunity_Data.csv
42.16 KB
-
README.md
8.82 KB
Abstract
Large ungulate populations can threaten forest regeneration and many rare or declining understory plants, birds, and small mammals. Reintroduction of large predators is often proposed as a remedy to reduce negative ecosystem effects associated with high ungulate populations, but we know little about the effectiveness of this approach. We assessed whether wolves (Canis lupus) can protect forest understory plants from excessive white-tailed deer (Odocoileus virginianus) browse. We planted white oak (Quercus alba), red oak (Quercus rubra), and zigzag goldenrod (Solidago flexicaulis) seedlings across a gradient of wolf density and occupancy time in northern Wisconsin and recorded deer browse and frequency of 11 common understory plants at each site. We found that wolf density and residence time had negative effects on deer browse intensity, but these effects were generally weak except when understory vegetation was abundant. Additionally, the presence of common understory plants decreased as a function of wolf density, opposite to what we would expect for a wolf-driven trophic cascade. The weak reduction in browse probability that is associated with wolves, particularly when vegetation is scarce, is unlikely to improve forest regeneration and recovery of understory plant communities currently threatened by high deer populations.
Dataset DOI: 10.5061/dryad.905qfttwv
Description of the data and file structure
This dataset contains the data and code required to replicate analyses in Brice et al. (in review), testing the hypothesis that wolves decrease deer browsing and subsequently increase forest regeneration. Data cover 42 sites in northern Wisconsin across a gradient of wolf density and wolf residence time, at which we planted white oak (Quercus alba), red oak (Quercus rubra), and zigzag goldenrod (Solidago flexicaulis) seedlings in the spring of 2022 and 2023. We monitored the seedlings for deer herbivory over 3 summers (2022-2024), and collected data on height growth and surrounding site conditions (e.g., vegetation height, leaf litter depth, canopy cover). During the 2022 field season, we also collected data on 11 common understory plants at each site. Files also include data on wolf density and residence time and site conditions (e.g., vegetation height, leaf litter depth, precipitation, tree size), and a Markdown file with code needed to replicate the analysis. Wolf density was estimated for the winter of 2020-2021 across 100-km2 hexagonal cells by Wisconsin Department of Natural Resources (WDNR) staff using a scaled occupancy model built using data derived from winter track surveys and GPS-collared wolves. We derived the number of years a site was occupied by wolves (i.e., wolf residence time) from overlapping annual maps of wolf pack locations from 1980, 1982, 1983, 1986, 1987, 1991-2010, and 2015-2019. These maps were created by WDNR by incorporating ground-based snow tracking surveys, aerial observations, and location data from radio-collared wolves to delineate pack areas and estimate pack size
Files and variables
File: Analysis_Code.Rmd
Description: An R Markdown file of code needed to replicate all analyses.
File: Covariates.csv
Description: site-level variables used in analyses
Variables
- Site: number used to identify the site
- Wolf_Den: wolf density at each site (wolves per 100 km2)
- Wolf_Time: number of years a wolf pack was present at the site
- Worms: presence of invasive earthworms at the site
- Litter_cm: average leaf litter depth at each site, measured at 4 points around each seedling and then averaged across all seedlings (cm)
- Precip_mm: previous summer precipitation as average monthly precipitation from May – September at each site (mm)
- Veg_cover: average vegetation cover (%) at each site, visually estimated and averaged at four points within 1-m of the seedling and then averaged across all seedlings.
- Veg_height: average vegetation height (cm) at each site, averaged at four points within 1-m of the seedling and then averaged across all seedlings. Height was measured up to 1-m.
- QMD_cm: quadratic mean diameter (cm) of trees at each site.
File: Browse_Data_Goldenrod.csv
Description: data used in the analysis of goldenrod browsing
Variables
- Site: number used to identify the site
- Tag_ID: ID number of individual seedling
- Height_1: height at planting (cm)
- Browse: binary variable for whether the seedling was browsed by deer
- Wolf_Den: wolf density at each site (wolves per 100 km2)
- Wolf_Time: number of years a wolf pack was present at the site
- Worms: presence of nonnative earthworms at the site
- Veg_height: vegetation height (cm) averaged at four points within 1-m of the seedling. Height was measured up to 1-m.
- Veg_cover: vegetation cover (%) visually estimated and averaged at four points within 1-m of the seedling.
- Litter: average leaf litter depth measured at 4 points around each seedling.
- Precip: previous summer precipitation as average monthly precipitation from May – September at each site (mm)
- QMD_cm: quadratic mean diameter (cm) of trees at each site.
File: Browse_Data_Oaks.csv
Description: data used in the analysis of oak browsing
Variables
- Site: number used to identify the site
- Transect: A indicates oaks planted in 2022, B indicates oaks planted in 2023.
- Tag_ID: ID number of individual seedling
- Species: oak species of seedling
- Height_1: height at planting (cm)
- Browse: binary variable for whether the seedling was browsed by deer
- Wolf_Den: wolf density at each site (wolves per 100 km2)
- Wolf_Time: number of years a wolf pack was present at the site
- Worms: presence of invasive earthworms at the site
- Veg_height: vegetation height (cm) averaged at four points within 1-m of the seedling. Height was measured up to 1-m.
- Veg_cover: vegetation cover (%) visually estimated and averaged at four points within 1-m of the seedling.
- Litter: average leaf litter depth measured at 4 points around each seedling.
- Precip: previous summer precipitation as average monthly precipitation from May – September at each site (mm)
- QMD_cm: quadratic mean diameter (cm) of trees at each site.
File: Growth_Data_Goldenrod.csv
Description: data used in the analysis of goldenrod growth
Variables
- Site: number used to identify the site
- Tag_ID: ID number of individual seedling
- Species: species of seedling (all goldenrod)
- Growth: change in height from May 2023 to August 2023
- Height_1: height (cm) at planting (May 2023)
- Height_2: height (cm) in August 2023
- Browse: binary variable for whether the seedling was browsed by deer
- Wolf_Den: wolf density at each site (wolves per 100 km2)
- Wolf_Time: number of years a wolf pack was present at the site
- Veg_height: vegetation height (cm) averaged at four points within 1-m of the seedling. Height was measured up to 1-m.
- Veg_cover: vegetation cover (%) visually estimated and averaged at four points within 1-m of the seedling.
- Litter: average leaf litter depth measured at 4 points around each seedling.
- Precip: previous summer precipitation as average monthly precipitation from May – September at each site (mm)
- QMD_cm: quadratic mean diameter (cm) of trees at each site.
File: Growth_Data_Oak.csv
Description: data used in the analysis of oak growth
Variables
- Site: number used to identify the site
- Tag_ID: ID number of individual seedling
- Species: oak species of the seedling
- Growth: change in height from June 2022 to August 2024
- Height_1: height (cm) at planting (June 2022)
- Height_6: height at visit 6 (August 2024)
- Browse: binary variable for whether the seedling was browsed by deer
- Other_damage: a binary predictor for other plant damage (e.g., rodent cuts, tip death, unknown damage)
- Wolf_Den: wolf density at each site (wolves per 100 km2)
- Wolf_Time: number of years a wolf pack was present at the site
- Worms: presence of invasive earthworms at the site
- Veg_height: vegetation height (cm) averaged at four points within 1-m of the seedling. Height was measured up to 1-m.
- Veg_cover: vegetation cover (%) visually estimated and averaged at four points within 1-m of the seedling.
- Litter: average leaf litter depth measured at 4 points around each seedling.
- Precip: previous summer precipitation as average monthly precipitation from May – September at each site (mm)
- QMD_cm: quadratic mean diameter (cm) of trees at each site.
File: PlantCommunity_Data.csv
Description: data used in the analysis of the plant community
Variables
- Site: number used to identify the site
- Species: species
- Freq: number of times that species was recorded at a site
- N: number of times we checked for species at a site
- Wolf_Den: wolf density at each site (wolves per 100 km2)
- Wolf_Time: number of years a wolf pack was present at the site
- Worms: presence of invasive earthworms at the site
- Veg_height: vegetation height (cm) averaged at four points within 1-m of the seedling, and averaged across all seedlings at a site. Height was measured up to 1-m.
- Veg_cover: vegetation cover (%) visually estimated and averaged at four points within 1-m of the seedling, and averaged across all seedlings at a site.
- Litter_site: average leaf litter depth at each site, measured at 4 points around each seedling and then averaged across all seedlings.
- Precip_Avg: previous summer precipitation as average monthly precipitation from May – September at each site (mm)
- Browse: the proportion of sentinel individuals browsed at each site
- QMD_cm: quadratic mean diameter (cm) of trees at each site.
Code/software
We used R v4.3.1 for all analyses, loaded with the following packages: tidyverse, lme4, lmertest, AICcmodavg, emmeans, gridExtra, DHARMa, car, and lmtest. The attached Markdown file includes the code needed to replicate the analysis.