Replication materials — High-resolution assessment of human pressure within Brazilian protected areas using the 10-metre Human Footprint Index

# Replication materials — High-resolution assessment of human pressure within Brazilian protected areas using the 10-metre Human Footprint Index This repository contains the **derived datasets, tabular outputs and analysis scripts** that support the manuscript: > Goycochea Casas G. (in revision) *High-resolution assessment of human > pressure within Brazilian protected areas using the 10-metre Human Footprint > Index*. Submitted to *Biodiversity and Conservation* (Springer). Once the article is accepted, this archive will be referenced in the manuscript's Data and Code Availability statements with a persistent DOI. --- ## Contents ``` repository_deposit/ ├── README.md this file ├── LICENSE.txt CC BY 4.0 (data) + MIT (code) — see file ├── requirements.txt Python dependencies for the analysis scripts ├── data/ derived datasets produced by the analysis ├── tables/ CSV source data for the manuscript tables └── scripts/ ordered analysis pipeline (01 → 07) └── sensitivity_resolution/ spatial-resolution sensitivity analysis ``` ### `data/` — derived datasets | File | Description | |------|-------------| | `wdpa_with_hfi.gpkg` | Per-protected-area (PA) GeoPackage with cleaned CNUC geometry attributes (area, perimeter, compactness, IUCN category, IUCN group, biome, size class) and the zonal HFI statistics (mean, median, max, SD, P90, CV, %HFI thresholds, %zero, etc.). `n` = 1,469 features; the 1,452 with valid HFI values are those analysed in the manuscript. CRS: EPSG:5880. | | `zonal_stats_raw.csv` | Long-form per-PA zonal statistics, keyed by `wdpa_id`. | | `gradient_profiles.csv` | Border-to-interior gradient profiles for 299 sampled PAs at six interior buffer distances (0, 100, 250, 500, 1000, 2000 m) and two exterior buffers (1 km, 5 km). | | `gradient_parameters.csv` | Per-PA decay-fit parameters (α, β, γ), penetration depth `d_50_m`, and fit R² for the same 299 PAs. | | `M2_inside_outside_by_biome.csv` | Per-biome median HFI inside vs outside PAs for the inside–outside contrast analysis (n=215 PAs with valid interior and exterior data). | | `M4_biome_overlap_comparison.csv` | Per-PA biome assignment under the centroid criterion vs dominant-area-overlap criterion, with the percentage of each PA inside its dominant biome (used for the biome-assignment robustness check in Methods). | | `M5_decayfit_subset.csv` | Per-PA flag indicating whether the exponential decay model yielded a usable `d_50` (subset of 95 PAs underlying Table 4 and the d₅₀ statistics). | ### `tables/` — manuscript tables and ancillary descriptives `Table1_sample_characterisation.csv` … `Table5_most_compromised.csv` correspond one-to-one with the tables printed in the manuscript. The remaining CSVs are ancillary descriptive outputs (per-biome, per-IUCN, per-size-class summaries; Kruskal–Wallis and Dunn post-hoc tables; the Spearman correlation matrix; ANOVA Type II partitioning; the LMM coefficient table; the interior–exterior contrast; etc.). These are the raw outputs used to populate the published tables and supporting text. ### `scripts/` — analysis pipeline The scripts are numbered to be run in order. They expect the layout of the original analysis project (see `config.py`); set the `ROOT` path in `config.py` to the local working directory before running. | Script | Purpose | |--------|---------| | `config.py` | Paths, constants (e.g., minimum PA area, HFI thresholds, biome palette). | | `01_prepare_data.py` | Downloads/prepares raw inputs (CNUC, IBGE biomes/states). | | `02_process_wdpa.py` | Cleans PA geometries, reprojects to EPSG:5880, computes area/perimeter/compactness, maps CNUC→IUCN, assigns biome by centroid, classifies size class. | | `03_zonal_stats_v2.py` | Extracts HIBR-10 zonal statistics for each PA (clip-and-stat for moderate PAs; random-point sampling for PAs > 50,000 km²). | | `04_gradient_v2.py` | Computes border-to-interior gradient profiles for a stratified random sample of ~300 PAs and fits the three-parameter exponential decay model. | | `05_statistical_analysis.py` | Non-parametric group comparisons (Kruskal–Wallis, Dunn), linear mixed-effects model (LMM), Type II ANOVA, Spearman correlation matrix. | | `06_figures_v2.py` | Generates Figs. 1–7 of the manuscript. | | `07_tables.py` | Generates the CSVs in `tables/`. | | `sensitivity_resolution/01_resolution_sensitivity.py` | Aggregates HIBR-10 from 10 m to 100 m, 300 m, 1 km and re-runs the PA-level zonal and gradient analyses. | ### Raw inputs (not redistributed here) The two raw inputs are public and should be obtained from their original sources: 1. **HIBR-10 (Human Footprint Index for Brazil, 10 m).** Presotto, Hamilton & Izar (2025). *A 10-meter resolution human footprint dataset to support biodiversity and conservation studies in Brazil*. Scientific Data **12**:1754. <https://doi.org/10.1038/s41597-025-06034-0> 2. **CNUC protected-area boundaries and IBGE biome/state layers**, accessed programmatically through the [`geobr`](https://ipeagit.github.io/geobr/) R/Python library; we used v0.2.9 in the analysis. --- ## Reproducing the analysis 1. Install the Python dependencies listed in `requirements.txt` (Python 3.12 recommended). 2. Edit `scripts/config.py` so that `ROOT` points to a local working directory that mirrors the expected structure (`ROOT/data/raw`, `ROOT/data/processed`, `ROOT/figures`, `ROOT/tables`, `ROOT/results`, `ROOT/logs`). 3. Place the raw inputs in `ROOT/data/raw/hfi/` (HIBR-10 GeoTIFF), `ROOT/data/raw/wdpa/` (CNUC GeoPackage) and `ROOT/data/raw/ibge/` (biomes and states GeoPackage). 4. Run the scripts in order. All random operations use fixed seeds. The derived datasets and tables that the scripts produce are bit-identical to those provided in `data/` and `tables/`. --- ## License - **Data** (`data/`, `tables/`): Creative Commons Attribution 4.0 International (CC BY 4.0). You are free to share and adapt with appropriate citation of the manuscript above. - **Code** (`scripts/`): MIT License. See `LICENSE.txt` for the full text of both licenses. ## Citation If you reuse any of these materials, please cite: > Goycochea Casas G. (in revision) High-resolution assessment of human > pressure within Brazilian protected areas using the 10-metre Human Footprint > Index. *Biodiversity and Conservation*. [DOI to be added on acceptance] and the HIBR-10 source dataset (Presotto et al. 2025, doi above). ## Contact Gianmarco Goycochea Casas — [e-mail] — ORCID [0000-0001-5491-8771](https://orcid.org/0000-0001-5491-8771). Department of Forest Engineering, Federal University of Viçosa, MG, Brazil.