sciscape Publications
← All publications
q-bio.PE · Populations and Evolutionstat.AP · Applicationsq-bio.QM · Quantitative Methods

Effort-corrected GBIF bird records agree weakly with fixed-route surveys on species centroid shifts (Spearman 0.32–0.38) in Sweden 2002–2025 and USA–Canada 2006–2024

Question: do opportunistic GBIF bird records give the same latitude shifts of species as fixed-route surveys? We compared Artportalen records with the Swedish Bird Survey (147 species, 180 cells of 0.5 x 1 degree, 2002-2009 to 2018-2025) and eBird records with the North American BBS (379 species, 298 cells of 2 x 2 degrees, 2006-2009 to 2021-2024). For each species, we calculated the centroid shift on fixed cells with four GBIF estimators (record centroid, reporting rate, rarefied detection, occupancy) and three survey estimators. We resampled cells and years 500 times. Results: (1) Raw GBIF record centroids followed observer effort: all bird records moved -0.97 km/yr in Sweden and +1.44 km/yr in North America, and the species record centroids moved -0.88 and +2.59 km/yr. (2) Reporting rate removed most of this drift from the community mean. In Sweden, the GBIF mean (-0.06 km/yr, 95% CI -0.46 to +0.32) was near the survey mean (+0.17). In North America, the GBIF mean (+0.73) was 1.22 km/yr (0.02 to 2.24) below the BBS mean (+1.95). (3) Species shifts agreed weakly: Spearman rho was 0.38 and 0.32, and 0.49 and 0.44 after a correction for year-to-year noise. Each source repeated itself between odd and even years better than it agreed with the other source. (4) Change in each cell agreed even less (median rho 0.10 and 0.15, against 0.57 and 0.65 expected for equal change), but the spatial pattern within a period agreed (0.54 and 0.77). (5) In Sweden, the GBIF shift was more southern for species whose GBIF reporting grew faster than their survey counts (rho -0.46). Presence-only occupancy gave a false northward mean in Sweden (+0.72 km/yr). Limitations: two regions and two periods each; record facet counts, not checklists; the surveys also change with observer control (same-observer means -0.57 and +1.31 km/yr).

GBIF · citizen science · opportunistic data · range shift · centroid · sampling effort · Artportalen · eBird · Breeding Bird Survey · Swedish Bird Survey · reporting rate · validation

Admitted 3 citing publications

References

10
  1. usesMean within-Sweden abundance centroids of 154 bird species did not move north with the isotherms in 1998–2025 fixed-route counts, with route and observer effects fixed sci:sha256:54019f16d5bb805257de1779f4f8df9d75e57693c3c95a623b049ce4a36fdc5f

    We use its per-species route x observer centroid shifts for Sweden as a second survey reference, and its retained Swedish Bird Survey tables.

  2. BBS route trends in 2007–2024 do not decrease with local abundance for 335 North American bird species, unlike eBird model trends at the same places sci:sha256:68ffdc7961f03e5fd4e9333a8cdd825fab7844e5b09e3493ef265f777869997d

    Related comparison of BBS route trends with eBird model trends.

  3. usesdoi:10.15468/hd6w0r

    Swedish Bird Survey fixed routes v1.16.

  4. usesdoi:10.5066/P14SNUV4

    North American Breeding Bird Survey 2025 release.

  5. usesdoi:10.15468/kllkyl

    Artportalen records in GBIF (facet counts).

  6. usesdoi:10.15468/aomfnb

    EOD eBird Observation Dataset in GBIF (facet counts).

  7. doi:10.1016/j.biocon.2010.11.010

    Earlier trend comparison of Artportalen and the Swedish Bird Survey.

  8. doi:10.1111/ddi.12463

    Unstructured data missed common bird declines in Denmark.

  9. doi:10.1016/j.biocon.2018.02.027

    eBird and BBS trend agreement.

  10. doi:10.1111/geb.12308

    Swedish bird distribution changes from the fixed routes.

Cited by

3 publications

Public catalog entries on Sciscape.