Data from: Passive acoustic monitoring captures large-scale Dryocopus pileatus (Pileated Woodpecker) occupancy patterns in the Pacific Northwest, USA

This repository contains the necessary data and code to reproduce the multi-state occupancy analysis featured in our paper, "Passive acoustic monitoring captures large-scale Dryocopus pileatus (Pileated woodpecker) occupancy patterns in the Pacific Northwest, USA," currently in prep. This description will be updated with a link to the paper upon successful publication. The repository is structured as an R project with a /data folder containing tabular data and a shapefiles subfolder containing one GIS shapefile necessary to complete the analysis. There is a main script called main.r which contains a small amount of contextual information. Running the main.r script will run the other four scripts in order, generating several output directories and a number of output files containing the main results of the analysis. The analysis was conducted in R (version 4.4.1) using several third-party packages, most notably RPresence (2.15.8), the tidyverse family of packages (release 2.0.0), and sf (1.0-20). The raw data are contained in the file data/PIWO_Encounter_Histories_w_Covariates_Raw.csv, which includes one line for each of 3,948 field sites (recording stations) where we placed autonomous recording units (ARUs) in Mar-Oct of 2023. We processed ca. 1.7 million hours of audio from these stations using the PNW-Cnet v5 deep learning model to detect Pileated Woodpecker vocalizations, treating clips with score >= 0.95 for the HYPI1 class as detections. We constructed encounter histories using detections from the first 6 weeks of recordings after manually reviewing detections from stations that had The analysis uses a fitted multi-state occupancy model to predict occupancy state across a set of 5-km^2 hexagons within the area monitored under the Northwest Forest Plan; this area includes the western portions of the US states of Washington and Oregon and the northwestern portion of California. The data file called data/Hex_Centroid_Covariate_Values_Raw.csv contains covariate values as measured at the center of each hexagon. Predictions are made only for hexagons that were at least 50% forested or forest-capable (n ~ 38,000); others are shown as NA in the table of predicted values. The data/shapefiles folder contains a shapefile listing all ~47,000 hexagons within the NWFP footprint and lists the percent federal ownership (Pct_Fed) and percentage of forest-capable cover (For_Cap) for each.