# Input data — Sarantaporos freshwater fish connectivity workflow These are the files the workflow **cannot generate**. Everything else in the analysis is produced by running the code. **Unzip this archive so its folders sit at the root of your data directory** (the one the workflow points at via the `WORKFLOW_DATA` environment variable, or `BASE_DIR` in `workflows/helpers/config.R`). Paths below are exactly the paths the scripts expect — do not rename or reorganise them. Total: 12 files, 2.4 MB. Checksums in `SHA256SUMS.txt` (`sha256sum -c SHA256SUMS.txt`). --- ## Fish occurrence and trait data — `points_original/fish/` | File | Used by | What it is | |---|---|---| | `fish_distributional_and_traits_data.xlsx` | Modules 1, 7, 10, 11 | Field survey occurrences (sheet `Data`) and species traits (sheet `Traits`). Sheet `legend` documents the columns. **The `Traits` sheet is the expert-compiled trait dataset for Greek freshwater fishes used throughout the analysis** — Module 10 reads it directly. A query to freshwaterecology.info returned no usable trait records for these species, so `10_traits/01_download_traits.R` is provided as a fallback for users who lack their own trait data, and its output is not used here. | | `Sarantaporos.xlsx` | Module 1 | Sarantaporos sub-basin survey records. | | `species_list_sarantaporos.txt` | Modules 7, 8 | The seven target species; drives the per-species loops. | | `fish_data_gbif.csv` | Module 1 (`03_clean_gbif_fish.R`) | The GBIF occurrence records used in this study (164 records, downloaded April 2026). See the GBIF note below. | | `gbif_citation.txt` | — | DOI of the **last** download only (the script overwrote this file per download); use `gbif_download_dois.tsv` instead. | | `gbif_raw_downloads_april2026/` | — | The original GBIF download archives and the full citation tables — see below. | ### The GBIF downloads, in detail The occurrence data was assembled from **13 separate GBIF downloads** made on 2026-04-29 (the script queries per taxonomic group). Their record counts sum to exactly the 164 records in `fish_data_gbif.csv`. `gbif_raw_downloads_april2026/` contains: | File | What it is | |---|---| | `gbif_download_dois.tsv` | **The citation list.** All 13 download keys with their DOIs and record counts, retrieved from the GBIF API on 2026-07-28. | | `gbif_datasets_all_downloads.tsv` | **The attribution list.** The 9 distinct source datasets across all 13 downloads, with dataset DOIs, record counts and full citation strings. | | `*.zip` (20 files) | The raw download archives. **Only 13 are part of this study** — those listed in `gbif_download_dois.tsv`. The other 7 (keys `0016386`, `0016387`, `0016388`, `0016403`, `0016404`, `0016407`, `0016422`, timestamped 12:45–14:00) are from an earlier attempt that was superseded by the 14:52 re-run; they are retained for completeness but did not contribute to `fish_data_gbif.csv`. | | `datasets_download_usage_0016812-...tsv` | GBIF's dataset-usage export for download `0016812` only (3 datasets, 21 records). Superseded by `gbif_datasets_all_downloads.tsv`, which covers all 13. | The 9 source datasets, by contribution: ZFMK Ichthyology collection (67), Freshwater fishes of Greece (35), INSDC Sequences (32), iNaturalist Research-grade Observations (16), International Barcode of Life / iBOL (8), Zoological collection of the National Museum (2), NABU|naturgucker (2), Observation.org (1), Classic localities of fishes from the Western Balkans (1). ### Why the GBIF data is deposited rather than re-downloaded `01_biodiversity_data/02_download_gbif_fish.R` can re-query GBIF, but a fresh query will **not** return this dataset: GBIF is continuously updated, and GBIF retains a download (and its DOI) for a limited period — the April 2026 download behind these results is scheduled for deletion. The occurrence records used in the analysis are therefore deposited here so the study stays reproducible after the DOI stops resolving. Re-running the download script reproduces the *procedure*, not the *dataset*. When citing, cite both the GBIF download DOI in `gbif_citation.txt` and the constituent datasets listed in the `.tsv` (ZFMK Ichthyology collection; INSDC Sequences; Freshwater fishes of Greece). ## Barrier data — `points_original/dams/` Hydropower register extracts for Greece. `01_clean_dam_data.R` maps licence stage onto the two scenarios: `HYDRO13_Operational` → existing dam, the remainder → planned. | File | What it is | |---|---| | `V_SDI_R_HYDRO13_Operational_Licence.csv` | Operational licences (the one existing dam) | | `V_SDI_R_HYDRO11_Production_Licence.csv` | Production licences | | `V_SDI_R_HYDRO12_Installation_Licence.csv` | Installation licences | | `V_SDI_R_HYDRO7_Evaluation.csv` | Applications under evaluation | | `V_SDI_R_HYDRO_OTHER_VALUES_Rejected.csv` | Rejected applications (excluded from scenarios) | | `dams_sarantaporos_table.csv` | Sub-basin dam table with plant names and capacity (MW); semicolon-delimited | ## Traits — `traits/` | File | Used by | What it is | |---|---|---| | `AspectRatioData.csv` | Module 7 (`07b_dispersal_distance.R`) | Caudal-fin aspect ratios; calibrates the `fishmove` dispersal model | | `traitsdata_freshwaterecology.csv` | — (not read by any script) | Trait table from freshwaterecology.info, retained for reference. The analysis uses the expert `Traits` sheet instead (see above). **See licensing note below.** | --- # NOT included — get these by running the code | What | How | |---|---| | GBIF occurrence records for *a new study area* | `01_biodiversity_data/02_download_gbif_fish.R` (GBIF API; needs credentials in `.Renviron`). The records for *this* study are deposited here — see above. | | Hydrography90m + CHELSA + ESA land cover tiles | `07_sdm/01_download_env90m_data.R` and `12_lakes/01_download_lake_data.R` (~tens of GB) | | `env90m/env_space_table.csv` | Module 6; the generating steps are commented out for speed — re-enable them once | | FishBase traits | `10_traits/01_download_traits.R` (API key in `.Renviron`) | | `spatial/hfp_zonal_stats.csv` (Human Footprint cost layer, Module 9) | `09_spatial_prioritization/00_human_footprint_cost_layer.R` — **but see the shortcut below, which avoids a 13 GB download** | ## Human Footprint layer — cropped rasters included as a shortcut `00_human_footprint_cost_layer.R` builds the Module 9 cost layer in four steps: 1. download the global Human Footprint COG (`hfp_2021_100m_v1-2_cog.tif`, **~13 GB**; source.coop / vizzuality) 2. crop it to the sub-basin → `spatial/hfp_crop.tif` 3. reproject to EPSG:4326 → `spatial/hfp_wgs.tif` 4. crop the `sub_catchment` tile → `spatial/subcatchment_sarantaporos.tif`, then `extract_zonal_stat()` → `spatial/hfp_zonal_stats.csv` The final CSV is a workflow product and is not shipped. **But the cropped intermediates are**, because they are the entire study-area extract of that 13 GB global raster and together weigh under 1 MB: | File | Size | Lets you skip | |---|---|---| | `spatial/hfp_crop.tif` | 156 KB | Steps 1–2 — the 13 GB download and the crop | | `spatial/hfp_wgs.tif` | 468 KB | Step 3 as well — reprojection | | `spatial/subcatchment_sarantaporos.tif` | 92 KB | The sub-catchment crop in step 4 | With these three in place you can run **step 4 alone** to regenerate `hfp_zonal_stats.csv` and reproduce the prioritisation, without downloading the global raster at all. The `.aux.xml` files alongside them are GDAL statistics caches and can be ignored or deleted. Values in `hfp_crop.tif` / `hfp_wgs.tif` are the Human Footprint Index on its native scale (roughly 170–25,400 here); `01_spatial_prioritization.R` rescales the zonal means to 0–1 itself. Around 40 % of cells are NA, which is expected — these are rectangular crops around a non-rectangular basin. To rebuild from the original global raster instead, re-enable the `download.file()` call at the top of the script (the URL is there) and run all four steps. Note the `00_` prefix: despite sitting in Module 9, this script must run **before** `01_spatial_prioritization.R`, which fails without its output. **SWOT Prior Lake Database** (`lakes/swot_lakes/swot_lakes.gpkg`, ~907 MB) is a manual download from . It is a third-party public database, so it is referenced rather than redistributed here. Module 12 expects it at that path.