{
  "dataset_key": "calpoly_whale-edna",
  "provider": {
    "key": "calpoly",
    "short": "Cal Poly",
    "name": "California Polytechnic State University",
    "url": "https://www.calpoly.edu",
    "registered": true
  },
  "dataset": "whale-edna",
  "dataset_name": "Baleen Whale Density Prediction from Environmental DNA (eDNA)",
  "dataset_name_short": null,
  "category": {
    "name": "Genomics & eDNA",
    "realm": "bio",
    "icon": "cat-genomics",
    "order": 13,
    "registered": true
  },
  "visibility": "public",
  "description_md": "Baleen Whale Density Prediction from Environmental DNA (eDNA) contains processed amplicon sequence variant (ASV) abundance tables for three eDNA markers, 16S rRNA, 18S rRNA V4, and 18S rRNA V9, developed by Erin Satterthwaite, Trevor Ruiz, Nastassia Patin, and colleagues to test whether microbial and small zooplankton communities can predict the density of baleen whales in the southern California Current Ecosystem. The dataset covers CalCOFI cruises in the southern California Current Ecosystem and spans 2014 to 2020, combining eDNA data from the NOAA CalCOFI Genomics Program (NCOG) with line-transect distance-sampling density estimates for blue, fin, and humpback whales derived from CalCOFI whale sighting surveys. The data are analyzed using stability selection adapted to sparse partial least squares (sPLS) regression, a statistical method for identifying reliable microbial and zooplankton predictors of whale density within large sequencing datasets. Processed data includes ASV tables with taxonomic annotations, sample metadata, density estimates, and cross-validation partitions. This dataset supports research on eDNA-based indicators of marine mammal distribution and is intended for researchers studying whale ecology and biomonitoring methods in the California Current.",
  "keywords": [
    "Zooplankton Biomass and Diversity",
    "Phytoplankton Biomass and Diversity"
  ],
  "attribution": {
    "citation_main": null,
    "citation_others": [],
    "doi": null,
    "license": null,
    "license_name": null,
    "license_url": null,
    "acknowledgement": null,
    "contact": "truiz01@calpoly.edu",
    "pi_names": [],
    "creators": [
      {
        "name": "Trevor Ruiz",
        "organization": "Cal Poly",
        "orcid": null,
        "email": "truiz01@calpoly.edu",
        "role": "lead"
      }
    ],
    "funding": "ONR",
    "source_accessed": null
  },
  "links": {
    "calcofi_org": null,
    "data_source": "https://doi.org/10.5281/zenodo.19139338",
    "page": "https://calcofi.io/datasets/calpoly_whale-edna/"
  },
  "distributions": [
    {
      "kind": "source",
      "url": "https://doi.org/10.5281/zenodo.19139338",
      "portal": "other",
      "title": "source",
      "status": "external"
    }
  ],
  "status": {
    "stage": "archived",
    "priority": null,
    "owner": null,
    "next_step": "confirm dataset scope, license and DOI with the provider; no ingest planned yet. Zenodo DOI (baleen whale eDNA) -- new provider calpoly",
    "module": "NA- one dataset (BUT NOTE: we will likely have more eDNA data that we want part of this catalog in the future)",
    "priority_caloos": "Other",
    "gh_issue": null
  }
}