{
  "dataset_key": "sio_cetacean-edna",
  "provider": {
    "key": "sio",
    "short": "SIO",
    "name": "Scripps Institution of Oceanography, UC San Diego",
    "url": "https://scripps.ucsd.edu",
    "registered": true
  },
  "dataset": "cetacean-edna",
  "dataset_name": "CalCOFI Cetacean eDNA Detections (NCOG samples)",
  "dataset_name_short": "Cetacean eDNA",
  "category": {
    "name": "Genomics & eDNA",
    "realm": "bio",
    "icon": "cat-genomics",
    "order": 13,
    "description": "Metabarcoding and environmental DNA — ASV tables, eDNA-derived abundance and density",
    "registered": true
  },
  "color": "#8a6bb0",
  "visibility": "public",
  "description_md": "Cetacean species detections from environmental DNA (eDNA) in seawater samples collected by the NOAA CalCOFI Genomics Program (NCOG) at 10-40 m (a few deeper) on CalCOFI cruises, 2014-2016. Seven dolphin and whale species detected by metabarcoding: 1 = detected, 0 = not detected in a sequenced sample (a non-detection, not a confirmed absence). The 62 samples listed as \"not sequenced, PCR negative\" did not amplify, so they say nothing about any species: each stays a sample with no edna_presence rows (not a 0) until the provider says what a PCR negative means (Q04). Samples have no parent sample (each is a root event); the CalCOFI bottle each was filtered from is matched only as evidence of its time and cruise. Each sample is timed by its CTD-rosette bottle cast. Staged from the CalCOFI marine mammal Shiny app repository (CalCOFI/marmam-app).\n",
  "keywords": [
    "EARTH SCIENCE > BIOLOGICAL CLASSIFICATION > ANIMALS/VERTEBRATES > MAMMALS",
    "Marine Turtles, Birds, and Mammals"
  ],
  "attribution": {
    "citation_main": null,
    "citation_others": [],
    "doi": null,
    "license": "unknown",
    "license_name": "Not yet established",
    "license_url": null,
    "acknowledgement": null,
    "contact": null,
    "pi_names": [
      "Simone Baumann-Pickering",
      "Julie Dinasquet"
    ],
    "creators": [
      {
        "name": "Simone Baumann-Pickering",
        "organization": "UCSD SIO",
        "orcid": null,
        "email": null,
        "role": "lead"
      },
      {
        "name": "Julie Dinasquet",
        "organization": "UCSD SIO",
        "orcid": null,
        "email": null,
        "role": "lead"
      }
    ],
    "funding": null,
    "source_accessed": "2026-10-04"
  },
  "links": {
    "calcofi_org": null,
    "data_source": "https://github.com/CalCOFI/marmam-app/tree/main/data/edna_data",
    "workflow": "https://calcofi.io/workflows/ingest_sio_cetacean-edna.html",
    "page": "https://calcofi.io/datasets/sio_cetacean-edna/"
  },
  "coverage": {
    "realm": "bio",
    "temporal": "2014-01 to 2016-11",
    "year_min": 2014,
    "year_max": 2016,
    "years": [
      {
        "year": 2014,
        "n_obs": 161,
        "n_roots": 23
      },
      {
        "year": 2015,
        "n_obs": 154,
        "n_roots": 22
      },
      {
        "year": 2016,
        "n_obs": 182,
        "n_roots": 26
      }
    ],
    "spatial": "31.5–35.1°N, 122.5–117.3°W",
    "bbox": {
      "lat_min": 31.513,
      "lat_max": 35.0882,
      "lon_min": -122.5333,
      "lon_max": -117.3054
    },
    "bbox_robust": {
      "lat_min": 31.51304,
      "lat_max": 35.02158,
      "lon_min": -122.24608,
      "lon_max": -117.30538,
      "n_positions": 497
    },
    "months": [
      126,
      14,
      0,
      126,
      0,
      0,
      91,
      0,
      0,
      0,
      140,
      0
    ],
    "n_obs": 497,
    "n_roots": 71,
    "n_stations": 21,
    "n_variables": 1,
    "n_taxa": 7,
    "depth_min_m": 0,
    "depth_max_m": 515,
    "variables": [
      {
        "name": "edna_presence",
        "units": "dimensionless",
        "uri": null,
        "category": "Genomics & eDNA"
      }
    ],
    "taxa": [
      {
        "taxon_key": "worms:137092",
        "scientific_name": "Megaptera novaeangliae",
        "common_name": "Humpback Whale",
        "rank": "Species",
        "n_obs": 71
      },
      {
        "taxon_key": "worms:137094",
        "scientific_name": "Delphinus delphis",
        "common_name": "Short-Beaked Common Dolphin",
        "rank": "Species",
        "n_obs": 71
      },
      {
        "taxon_key": "worms:137098",
        "scientific_name": "Grampus griseus",
        "common_name": "Risso's Dolphin",
        "rank": "Species",
        "n_obs": 71
      },
      {
        "taxon_key": "worms:137111",
        "scientific_name": "Tursiops truncatus",
        "common_name": "Bottlenose Dolphin",
        "rank": "Species",
        "n_obs": 71
      },
      {
        "taxon_key": "worms:1571909",
        "scientific_name": "Sagmatias obliquidens",
        "common_name": "Pacific White-Sided Dolphin",
        "rank": "Species",
        "n_obs": 71
      },
      {
        "taxon_key": "worms:254975",
        "scientific_name": "Lissodelphis borealis",
        "common_name": "Northern Right Whale Dolphin",
        "rank": "Species",
        "n_obs": 71
      },
      {
        "taxon_key": "worms:383816",
        "scientific_name": "Delphinus capensis",
        "common_name": "Long-beaked Common Dolphin",
        "rank": "Species",
        "n_obs": 71
      }
    ],
    "life_stages": [],
    "contributes_to": []
  },
  "tables": [
    "sample",
    "obs",
    "taxon",
    "dataset_taxon",
    "taxon_group",
    "measurement_type"
  ],
  "objects": [
    {
      "table": "obs",
      "scope": "partition",
      "shared": false,
      "path": "ducklake/tables/obs/dataset_key=sio_cetacean-edna/ccdb631db5874f4e12a93f88/data_0.parquet",
      "url": "https://storage.googleapis.com/calcofi-db/ducklake/tables/obs/dataset_key=sio_cetacean-edna/ccdb631db5874f4e12a93f88/data_0.parquet",
      "table_description": "The occurrence-headline long table: one scalar measurement per row, environmental (`realm = 'env'`) and biological (`realm = 'bio'`) together.",
      "bytes": 7210,
      "sha256": "7d32c6f059672b2598fda1f15d9bfc22a787d7e86c6685382ba5aa829a953a07",
      "since": "v2026.10.05"
    }
  ],
  "since_version": "v2026.10.05",
  "distributions": [
    {
      "kind": "service",
      "url": "https://storage.googleapis.com/calcofi-db/stac/collections/sio_cetacean-edna/collection.json",
      "format": "stac",
      "id": "sio_cetacean-edna",
      "title": "STAC collection",
      "status": "current"
    },
    {
      "kind": "download",
      "url": "https://storage.googleapis.com/calcofi-db/ducklake/tables/obs/dataset_key=sio_cetacean-edna/ccdb631db5874f4e12a93f88/data_0.parquet",
      "format": "parquet",
      "table": "obs",
      "scope": "partition",
      "shared": false,
      "title": "obs (this dataset's rows)",
      "table_description": "The occurrence-headline long table: one scalar measurement per row, environmental (`realm = 'env'`) and biological (`realm = 'bio'`) together.",
      "bytes": 7210,
      "sha256": "7d32c6f059672b2598fda1f15d9bfc22a787d7e86c6685382ba5aa829a953a07",
      "since": "v2026.10.05",
      "status": "current"
    },
    {
      "kind": "notebook",
      "url": "https://calcofi.io/workflows/ingest_sio_cetacean-edna.html",
      "format": "html",
      "title": "ingest notebook",
      "status": "current"
    },
    {
      "kind": "source",
      "url": "https://github.com/CalCOFI/marmam-app/tree/main/data/edna_data",
      "portal": "other",
      "title": "source",
      "status": "external"
    }
  ],
  "registrations": [
    {
      "portal": "erddap",
      "status": "planned",
      "issues": []
    },
    {
      "portal": "obis",
      "status": "planned",
      "issues": []
    },
    {
      "portal": "edi",
      "status": "planned",
      "issues": []
    },
    {
      "portal": "ncei",
      "status": "n/a",
      "issues": []
    },
    {
      "portal": "caloos",
      "status": "n/a",
      "issues": []
    },
    {
      "portal": "zenodo",
      "status": "published",
      "url": "https://doi.org/10.5281/zenodo.22281994",
      "id": "10.5281/zenodo.22281994",
      "title": "CalCOFI Integrated Database (the release deposit)",
      "note": "as part of the integrated database release"
    }
  ],
  "status": {
    "stage": "validated",
    "priority": "should-complete",
    "gh_issue": null,
    "blockers": "Re-rendered and validated 2026-10-01; un-held 2026-10-04 (ships in the next release, licence pending). 133 NCOG samples, all timed and keyed by their calcofi_bottle cast (match_by_site_datetime, then a Niskin bottle within 4 m by match_nearest_by_depth), CC1611 included. 71 sequenced samples publish 497 edna_presence obs (15 detections); the 62 PCR-negative samples are held as sample rows with no obs until Q04 is answered. Dc keyed to Delphinus capensis; line/station from the sample name (Q10). Open: licence Q01, citation Q02, provider Q03, PCR negatives Q04.",
    "publish_policy": "Archive of record: none known; not the calpoly_whale-edna Zenodo deposit.",
    "updated": "2026-10-01",
    "questions_open": 10,
    "questions_dataset": [
      {
        "label": "Q01",
        "field": "license",
        "status": "open"
      },
      {
        "label": "Q02",
        "field": "citation_main",
        "status": "open"
      },
      {
        "label": "Q03",
        "field": "pi_names",
        "status": "proposed"
      }
    ]
  }
}