Data from: Using passive acoustic monitoring and LiDAR to conduct a statewide assessment of ruffed grouse (Bonasa umbellus) occurrence in Pennsylvania
Data files
Apr 09, 2026 version files 907.83 KB
-
Connectedness_focal_stats.R
2.94 KB
-
elev_focal_stats.R
2.94 KB
-
FIA_Conifer_focal_stats.R
3.49 KB
-
FIA_hardwoods_focal_stats.R
3.46 KB
-
LiDAR_RUGR_Top_combo.R
4.80 KB
-
long_focal_stats.R
2.37 KB
-
Master_Grouse_all_covs.csv
790.58 KB
-
NLCD_Forest_focal_stats.R
4.66 KB
-
p90_focal_stats.R
2.97 KB
-
pfr_focal_stats.R
2.94 KB
-
Predictive_Maps.R
9.13 KB
-
README.md
11.31 KB
-
RUGR_LiDAR_landscape_analysis.R
36.18 KB
-
RUGR_LiDAR_LiDAR_analysis.R
30.07 KB
Abstract
Effective conservation of wildlife is often hindered by poor understanding of where focal species and their habitats occur across large landscapes. Advancements in remote sensing have enabled researchers to improve detection of focal species (e.g., autonomous recording units; ARUs) and the characterization of their habitats (e.g., light detection and ranging [LiDAR]), thus mitigating these issues. Research into Ruffed Grouse (Bonasa umbellus), a declining forest game bird, stands to benefit from these technologies given the species’ low detectability and preference for particular forest structure conditions that are difficult to capture using imagery-based remotely sensed data. Herein, we investigated regional occurrence of Ruffed Grouse across Pennsylvania via the use of a multi-year passive acoustic monitoring dataset, fine scale LiDAR-derived forest structure metrics, and a suite of other forest and landscape variables to predict state-wide Ruffed Grouse occurrence probability and identify areas in need of targeted habitat management. Our analyses indicated that, well-connected, high-elevation hardwood forests with some conifers and well-developed understories (e.g., timber harvests) were predicted to have the highest probability of grouse occurrence. Likewise, similar forests with open understories (e.g., mature forests within similar landscape contexts) were predicted to be the most promising for future management. ARUs proved to be effective at building a large detection dataset, which when paired with the superior forest structure data provided by LiDAR, allowed us make predictions about Ruffed Grouse across Pennsylvania with greater confidence than ever before.
Dataset DOI: 10.5061/dryad.hmgqnk9xh
Dataset description
Autonomous recording units (ARUs) were deployed in the spring from 2020-2023 to capture ruffed grouse (bonasa umbellus) in Pennsylvania. Only locations that recorded for 5 days or more were included in the data set, and a single detection rendered a location "used" (1) while locations without any detections were considered "unused" (0). Various remotely sensed data, including LiDAR, NLCD, FIA forest type group, elevation, and aspect were extracted at several different radii at sampling locations.
Files
Master_Grouse_all_covs.csv
Description Detection data (response) and all site covariates that were used at predictor variables. Some columns are also included that were collection-related or for exploring the dataset, and so were not important to the analysis. Described below are the columns that appear in this file. Location data (lat/long) has been obscured to protect private landowner privacy and prevent the misuse of this data for hunting.
Variable Descriptions:
X.1: Ignore, data processing artifact.
X: Ignore, data processing artifact.
OID_: Ignore, data processing artifact.
point_ID: Unique identifier of sampling locations.
SD_id: Unique identifier of first SD card used in ARU.
SD_id_2: Unique identifier of second SD card used in an ARU sampling location that season if the first was swapped out.
year: Year of data collection for that sampling location.
project: Original research project for which ARUs were deployed.
lat: Latitude of sampling location in decimal degrees. Masked for landowner privacy and to prevent misuse of data.
long: Longitude of sampling location in decimal degrees. Masked for landowner privacy and to prevent misuse of data.
Any_Det_in_5_weeks_:Binary presence/absence for ruffed grouse for all sampled days per unit. 1 = grouse detected, 0 = no grouse detected.
days recorded: The number of calendar days that an ARU (autonomous recording unit) recorded audio at each site.
Detection_4_15_to_5_19: Total days in which a drumming ruffed grouse was detected within the survey window. Used to explore data but not used in analysis.
p95TR30_50m: Standard deviation of the height of the 95th percentile of returns (m) summarized at a 30 m radius for each pixel. Extracted at 50m radius.
p95TR30_100m: Standard deviation of the height of the 95th percentile of returns (m) summarized at a 30 m radius for each pixel. Extracted at 100m radius.
p95TR30_250m: Standard deviation of the height of the 95th percentile of returns (m) summarized at a 30 m radius for each pixel. Extracted at 250m radius.
p95TR30_500m: Standard deviation of the height of the 95th percentile of returns (m) summarized at a 30 m radius for each pixel. Extracted at 500m radius.
p75_50m: Height of 75th percentile of returns counting from the ground (m). Extracted at 50m radius.
p75_100m: Height of 75th percentile of returns counting from the ground (m). Extracted at 100m radius.
p75_250m: Height of 75th percentile of returns counting from the ground (m). Extracted at 250m radius.
p75_500m: Height of 75th percentile of returns counting from the ground (m). Extracted at 500m radius.
pfr_50m: Percent of first returns reflected at 1–5 m off of the ground. Extracted at 50m radius.
pfr_100m: Percent of first returns reflected at 1–5 m off of the ground. Extracted at 100m radius.
pfr_250m: Percent of first returns reflected at 1–5 m off of the ground. Extracted at 250m radius.
pfr_500m: Percent of first returns reflected at 1–5 m off of the ground. Extracted at 500m radius.
p90_50m: Height of 90th percentile of returns counting from the ground (m). Extracted at 50m radius.
p90_100m: Height of 90th percentile of returns counting from the ground (m). Extracted at 100m radius.
p90_250m: Height of 90th percentile of returns counting from the ground (m). Extracted at 250m radius.
p90_500m: Height of 90th percentile of returns counting from the ground (m). Extracted at 500m radius.
IQR_50m: Interquartile range of returns (m). Extracted at 50m radius.
IQR_100m: Interquartile range of returns (m). Extracted at 100m radius.
IQR_250m: Interquartile range of returns (m). Extracted at 250m radius.
IQR_500m: Interquartile range of returns (m). Extracted at 500m radius.
MOCHTR30_50m: Standard deviation of mean outer canopy height (m) summarized at a 30 m radius for each pixel. Extracted at 50m radius.
MOCHTR30_100m: Standard deviation of mean outer canopy height (m) summarized at a 30 m radius for each pixel. Extracted at 100m radius.
MOCHTR30_250m: Standard deviation of mean outer canopy height (m) summarized at a 30 m radius for each pixel. Extracted at 250m radius.
MOCHTR30_500m: Standard deviation of mean outer canopy height (m) summarized at a 30 m radius for each pixel. Extracted at 500m radius.
MOCHTR50_50m: Standard deviation of mean outer canopy height (m) summarized at a 50 m radius for each pixel. Extracted at 50m radius.
MOCHTR50_100m: Standard deviation of mean outer canopy height (m) summarized at a 50 m radius for each pixel. Extracted at 100m radius.
MOCHTR50_250m: Standard deviation of mean outer canopy height (m) summarized at a 50 m radius for each pixel. Extracted at 250m radius.
MOCHTR50_500m: Standard deviation of mean outer canopy height (m) summarized at a 50 m radius for each pixel. Extracted at 500m radius.
p95TR50_50m: Standard deviation of the height of the 95th percentile of returns (m) summarized at a 50m radius for each pixel. Extracted at 50m radius.
p95TR50_100m: Standard deviation of the height of the 95th percentile of returns (m) summarized at a 50m radius for each pixel. Extracted at 100m radius.
p95TR50_250m: Standard deviation of the height of the 95th percentile of returns (m) summarized at a 50m radius for each pixel. Extracted at 250m radius.
p95TR50_500m: Standard deviation of the height of the 95th percentile of returns (m) summarized at a 50m radius for each pixel. Extracted at 500m radius.
p99TR30_50m: Standard deviation of the height of the 99th percentile of returns (m) summarized at a 30m radius for each pixel. Extracted at 50m radius.
p99TR30_100m: Standard deviation of the height of the 99th percentile of returns (m) summarized at a 30m radius for each pixel. Extracted at 100m radius.
p99TR30_250m: Standard deviation of the height of the 99th percentile of returns (m) summarized at a 30m radius for each pixel. Extracted at 250m radius.
p99TR30_500m: Standard deviation of the height of the 99th percentile of returns (m) summarized at a 30m radius for each pixel. Extracted at 500m radius.
p99TR50_50m: Standard deviation of the height of the 99th percentile of returns (m) summarized at a 50m radius for each pixel. Extracted at 50m radius.
p99TR50_100m: Standard deviation of the height of the 99th percentile of returns (m) summarized at a 50m radius for each pixel. Extracted at 100m radius.
p99TR50_250m: Standard deviation of the height of the 99th percentile of returns (m) summarized at a 50m radius for each pixel. Extracted at 250m radius.
p99TR50_500m: Standard deviation of the height of the 99th percentile of returns (m) summarized at a 50m radius for each pixel. Extracted at 500m radius.
NLCD_Deciduous: Percent of 500m extraction radius classified as deciduous forest by NLCD.
NLCD_Evergreen: Percent of 500m extraction radius classified as evergreen forest by NLCD.
NLCD_Mixed: Percent of 500m extraction radius classified as mixed forest by NLCD.
NLCD_Forested: Percent of 500m extraction radius classified as any forest type by NLCD.
NLCD_Shrub_Scrub: Percent of 500m extraction radius classified as schrub scrub forest by NLCD.
elev: Elevation in meters at ARU sampling location.
slope: Percent slope at ARU sampling location.
aspect: Ground aspect degree of direction at ARU sampling location.
p_1to5_50m: Percent of LiDAR returns reflected at 1–5 m off of the ground. Extracted at 50m radius.
p_1to5_100m: Percent of LiDAR returns reflected at 1–5 m off of the ground. Extracted at 100m radius.
p_1to5_250m: Percent of LiDAR returns reflected at 1–5 m off of the ground. Extracted at 250m radius.
p_1to5_500m: Percent of LiDAR returns reflected at 1–5 m off of the ground. Extracted at 500m radius.
Use.Rate: Percentage of days recorded in which a drumming ruffed grouse was detected.
p99_50m: Height of 99th percentile of returns counting from the ground (m). Extracted at 50m radius.
p99_100m: Height of 99th percentile of returns counting from the ground (m). Extracted at 100m radius.
p99_250m: Height of 99th percentile of returns counting from the ground (m). Extracted at 250m radius.
p99_500m: Height of 99th percentile of returns counting from the ground (m). Extracted at 500m radius.
Connectivity_Level: Categorical level of connectedness from TNC's Resilient Connected Network Tool.
mean_connectedness: Index value of connectedness from TNC's Resilient Connected Network Tool.
Oak: % oak hickory and oak pine FIA forest type groups. Extracted at 500m radius.
N.Hardwoods: % maple beech birch or aspen birch FIA forest type groups. Extracted at 500m radius.
p_2to6_50m: Percent of returns reflected at 2–6 m off of the ground. Extracted at 50m radius.
p_2to6_100m: Percent of returns reflected at 2–6 m off of the ground. Extracted at 100m radius.
p_2to6_250m: Percent of returns reflected at 2–6 m off of the ground. Extracted at 250m radius.
p_2to6_500m: Percent of returns reflected at 2–6 m off of the ground. Extracted at 500m radius.
Conifer:: % white/red/jack pine, spruce/fir, or oak pine FIA forest type groups. Extracted at 500m radius.
Code/software
All code was run in R version 4.3.2. Files should be run in the order as the files appear below.
RUGR_LiDAR_landscape_analysis.R
Description Code for the landscape portion of the analysis.
RUGR_LiDAR_LiDAR_analysis.R
Description Code for the vegetative structure (LiDAR) portion of the analysis.
LiDAR_RUGR_Top_combo.R
Description Code for the combined model portion of the analysis.
Connectedness_focal_stats.R
elev_focal_stats.R
FIA_Conifer_focal_stats.R
FIA_hardwoods_focal_stats.R
long_focal_stats.R
NLCD_Forest_focal_stats.R
p90_focal_stats.R
pfr_focal_stats.R
Description Code in eight "focal stat" files follows the same format for each of the mappable covariates included in the top models/combined model. The purpose of this step was to prepare covariate-specific layers to be used in the predictive maps.
Predictive_Maps.R
Description Code for producing predictive maps.