This repository contains the public code and anonymized data objects needed to reproduce the analyses and non-spatial outputs from:

Fiss, C. J., et al. Automated fledgling call detection links adult songbird abundance to reproductive success in restored habitat.

Repository purpose

The goal of this repository is to provide a transparent, reproducible workflow for fitting the joint adult–fledgling abundance model used in the manuscript while protecting the exact locations of survey sites. To do that, all spatial coordinates are redacted (forced to 0) rather than real-world coordinates.

This repository reproduces:

• the fitted joint JAGS model
• posterior summaries
• Bayesian p-values
• Moran’s I tests on model residuals for demonstration only; site coordinates redacted
• coefficient plots for back-transformed landcover and groundcover PCA effects
• the 4-panel pathway / effect figure reported in the manuscript

This repository does not include raw field data processing, machine-learning classifier development, or raw site coordinates.

Repository structure

data_public/

Contains anonymized public data objects required to run the joint model and produce figures.

gwwa_joint_public_data.rds

R object containing the public data used in the model-fitting and output scripts. This object includes the site-level and observation-level inputs needed to fit the public JAGS model.

Expected contents include objects such as:

• adult_umf – adult detection unmarkedFrame
• fledge_umf – fledgling detection unmarkedFrame
• Preliminary detection-only AIC tables derived from frequentist style RoyleNichols models for both adults and fledglings (separately)

Variables Defined (same for adult_umf and fledge_umf):

• "y.1" - "y.10" = Adult/Fledgling detection histories (0/1) for 10 day period
• Year.x = Year of data collection
• Hab_type = main habitat type of site (aspen, hardwood, alder)
• Avg_prism = Average basal area reading from site using 10 factor cruising prism (scaled)
• Avg_LL = Average leaf litter groundcover estimate from site (scaled) 
• Avg_woody = Average woody vegetation groundcover estimate from site (scaled)
• Avg_grass = Average grass/graminoid groundcover estimate from site (scaled)
• Avg_forbs = Average forb groundcover estimate from site (scaled)
• Avg_rub = Average Rubus (spp.) groundcover estimate from site (scaled)
• Avg_fern = Average fern groundcover estimate from site (scaled)
• X.Open.Water = percent of 1km radius circle consisting of "open water" National Landcover Database (NLCD) classification (scaled)
• X.Developed..Open.Space = percent of 1km radius circle consisting of "developed open space" NLCD classification (scaled)
• X.Developed..Low.Intensity = percent of 1km radius circle consisting of "developed low intensity" NLCD classification (scaled)
• X. Developed..Medium.Intensity = percent of 1km radius circle consisting of "developed medium intensity" NLCD classification (scaled)
• X.Developed..High.Intensity = percent of 1km radius circle consisting of "developed high intensity" NLCD classification (scaled)
• X.Barren.Land = percent of 1km radius circle consisting of "barren land" NLCD classification (scaled)
• X.Deciduous.Forest = percent of 1km radius circle consisting of "deciduous forest" NLCD classification (scaled)
• X.Evergree.Forest = percent of 1km radius circle consisting of "evergreen forest" NLCD classification (scaled)
• X.Mixed.Forest = percent of 1km radius circle consisting of "mixed forest" NLCD classification (scaled)
• X.Shrub.Scrub = percent of 1km radius circle consisting of "shrub scrub" NLCD classification (scaled)
• X.Grassland.Herbaceous = percent of 1km radius circle consisting of "grassland/herbaceous" NLCD classification (scaled)
• X.Pasture.Hay = percent of 1km radius circle consisting of "pasture/hay" NLCD classification (scaled)
• X.Cultivated.Crops = percent of 1km radius circle consisting of "cultivated/crops" NLCD classification (scaled)
• X.Woody.Wetlands = percent of 1km radius circle consisting of "woody wetlands" NLCD classification (scaled)
• X.Emergent.Wetlands = percent of 1km radius circle consisting of "emergent wetlands" NLCD classification (scaled)
• aspen = binary (main habitat type = aspen)
• hardwood = binary (main habitat type = hardwood)
• alder = binary (main habitat type = alder)
• dev1 = sum of X.Developed..Open.Space, X.Developed..Low.Intensity, X.Developed..Medium.Intensity, X.Developed..High.Intensity
• ag = sum of X.Grassland.Herbaceous, X.Cultivated.Crops, X.Pasture.Hay
• X.Evermixed = sum of X.Evergreen.Forest, X.Mixed.Forest
• "day.1" - "day.10" = day of survey (ordinal)
• "noise.1" - "noise.10" = background noise captured on audio recording units' full frequency range across 1.5hr recording period for each survey day. Measured as Decibels Relative to Full Scale (dBFS).

Note: Spatial coordinates in this object are all set to 0 to protect private landowner privacy. All results will be slightly different than the published versions.

scripts/

1_fit_joint_models_public.R

Fits the joint adult–fledgling abundance model in JAGS using the anonymized public data object.

This script:

• loads the public data object
• defines the JAGS model
• runs the public model (mod2)
• saves the fitted model object for downstream summaries and figures

Expected output:

• output/jags_out_public.rds

2_make_outputs_figures_public.R

Generates the manuscript outputs that can be reproduced from the public model object and public data object.

This script produces:

• posterior summary table
• Bayesian p-values
• latent state summaries
• Moran’s I test code for demonstration only (coordinates redacted)
• back-transformed coefficient plots for PCA-based landcover and groundcover effects
• the 4-panel figure illustrating adult abundance, adult–fledgling relationships, pathway effects, and fledgling abundance contrasts

Expected outputs include files such as:

• output/posterior_summary.csv
• output/moran_results.txt
• output/latent_state_summaries.rds
• output/adult_fledge_lc_gc_coeff.jpg
• output/pathway_plot2.jpg

output/

This folder is created by the scripts and stores model outputs, summary tables,

and figures.

premade_output/

This folder contains fitted model objects for jags_mod1 and jags_mod2 created in script "1_fit_joint_models_public.R" so that, optionally, the jags fitting step can be skipped.

Recommended workflow

Run the scripts in the following order:

1) Fit the public model

Run:

source("scripts/1_fit_joint_models_public.R”) This will fit the public JAGS model and save the fitted model object to:

output/jags_out_public.rds

2) Generate outputs and figures

Run:

source("scripts/2_make_outputs_figures_public.R”) This script reads the public data object and fitted JAGS object, then reproduces the non-spatial outputs associated with the manuscript.

Data privacy and anonymization

Why are the coordinates redacted?

Survey locations used in this study are not included in their original form in order to protect site confidentiality and avoid publicly releasing coordinates associated with private landowner locations.

What is included instead?

The repository includes all data and code except coordinates to allow:

• rerunning the exact models used in analysis
• demonstration of Moran’s I tests of residual spatial autocorrelation in the public workflow

Important consequence

Because Moran’s I in the public repository is not based on coordinates, Moran’s . I values from the public workflow are irrelevant

The model structure, covariates, priors, and fitted public workflow are otherwise intended to match the manuscript analysis.

What is not included in this repository

This public repository is intended to reproduce the manuscript’s model fitting and similar outputs, not every upstream data-processing step used during project development. As a result, the following are not included:

• raw field data files
• survey coordinates
• intermediate private data wrangling scripts used to construct working objects
• machine-learning classifier training workflows for fledgling begging call detection (see Appendix A)
• mapping scripts based on coordinates
• any scripts whose only purpose was to prepare confidential data products prior to creation of the public model object

Instead, the repository begins from the public analysis-ready data object used

to fit the public model.

Software requirements

The analyses were run in R using JAGS for Bayesian model fitting.

Main R packages

This repository uses packages including:

dplyr tidyr ggplot2 patchwork spdep ggrepel jagsUI rjags unmarked

Reproducibility notes

Model object naming

The manuscript results and public outputs are based on model 2 (mod2) . The output scripts assume the fitted model object stored in

output/jags_out_public.rds corresponds to that model.

Moran’s I

Residual Moran’s I in the public repository is for demonstration of workflow only. Moran’s I results based on data in the public repository are irrelevant.

Figures

The public output script reproduces:

• coefficient plots for back-transformed PCA effects
• 4-panel figure associated with adult abundance, fledgling abundance, and pathway interpretation

Mapping products based on coordinates are intentionally excluded from the public repository.

Citation

If you use this repository, please cite the associated manuscript:

Fiss, C. J., et al. Automated fledgling call detection links adult songbird abundance to reproductive success in restored habitat.

• [Add journal, year, volume, pages, DOI once available]

Contact

For questions about the repository or requests related to the analysis workflow, please contact:

Cameron Fiss fisscameron@gmail.com