Polar bear at Churchill: what the map shows around it
At Churchill, on Hudson Bay, the tundra barely changes. What limits the polar bear is sea ice, which a land-cover map does not show.
How to cite: Biondo, Leandro M. (2026). Polar bear at Churchill: what the map shows around it. UmaTerra. https://umaterra.com.br/en/stories/endangered-species/polar-bear/
817records of the species on GBIF, within <span class="ea-n" data-r="272">50</span> km measured
961.4 haof tree-cover loss within <span class="ea-n" data-r="272">50</span> km (Hansen) measured
+2.9 °Cin mean temperature by mid-century, SSP2-4.5 scenario estimated
measured by satellite or by a GBIF count.
estimated by models (climate).
missing a part not computed yet.
1
Polar bear, Ursus maritimus
The polar bear (Ursus maritimus) is listed as Vulnerable (VU) on the IUCN Red List. At Churchill, the question is whether the tundra and the sea-ice coast changed in land cover, knowing from the start that the limit for the species is ice.
IUCN Red List category: Vulnerable (VU), assessed in 2015. The story's place is Churchill, at 58.77° N, 94.17° W.
2
Who the species is
In Western Hudson Bay (captures 1980-2019), apparent survival of prime-age males declined 10% during the earliest sea-ice retreat dates. Polar bears were listed under the US Endangered Species Act in 2008, mostly due to the threat from declining Arctic sea ice. Bear outcomes worsened over time to 2100 under both stabilised and unabated greenhouse-gas pathways.
The 6 pictures alongside are observations published on GBIF within 50 km of the place, each with its author's name and the licence they chose.
3
Where GBIF finds it
In a disc of 50 km radius around Churchill, GBIF holds 817 records of Ursus maritimus with coordinates and no known geospatial issue; 374 of them lie within 10 km of the point. By decade, 352 from the 2020s and 366 from the 2010s; the oldest dated one is from 1968. They are 652 human observations, 156 material samples and 9 preserved specimens; 580 human observations have a picture. They come from 6 datasets; the largest is iNaturalist Research-grade Observations, with 649. These numbers say where and when someone looked and published what they saw, not how many animals live there: GBIF measures observation effort, not abundance.
In the same disc, GBIF tags another 26 species with an IUCN threat category (2 critically endangered, 2 endangered and 22 vulnerable). The category is the one GBIF attaches to the name and may be out of date; the table shows the most recorded ones.
4
The place and its land cover
We found no verified source on land-use change at Churchill. The land-cover map shows the tundra and the coast, and the species' story is told by the ice.
The map used here is NTEMS VLCE2, with its classes gathered into common groups, from 1987 to 2022. In the 50 km disc, the three largest groups in the last year: wetland, from 233,568 ha (30%) to 251,723 ha (32%); water, from 238,648 ha (31%) to 239,146 ha (31%); other natural vegetation (grass, shrub, savanna), from 78,818 ha (10%) to 75,532 ha (10%). In the 10 km disc, closer to the point: forest and tree plantation from 2% to 3% of the mapped area. VLCE2 is a forested-ecosystem map: it has no cropland or built-up class.
5
Tree-cover loss and fire
Hansen counts 961.4 ha of tree-cover loss between 2001 and 2025 in the 50 km disc: 1% of the area that had at least 30% canopy in 2000 (162,433 ha). In the 10 km disc, 0.14 ha (0%). The year with most loss was 2013, with 275.0 ha. Loss counts any cause: clearing, fire, storm, plantation harvest. Hansen's gain has no date and stops in 2012, so it is not subtracted.
MODIS saw burned area in 8 of the 25 years with data in the 50 km disc; the largest yearly total was 1,994 ha, in 2007. Pixels are 500 m and small fires slip through; adding the years counts twice what burned again.
On the Canada page, the NTEMS forest-age map covers 15.7% of the 50 km disc in 2019, and the median age falls in the 61–80 years bin.
6
The country map and the world map
The two maps have different legends; to compare them, each map's classes were gathered into the same eight common groups and crossed pixel by pixel.
In 2022, NTEMS VLCE2 and GLC-FCS30D give the same common group on 52.5% of the area both cover in the 50 km disc. Forest on both maps: 42,037 ha; only on NTEMS VLCE2: 13,359 ha; only on GLC-FCS30D: 118,405 ha. The largest difference is in wetland: 251,723 ha on NTEMS VLCE2 and 100,621 ha on GLC-FCS30D. VLCE2 has no cropland or built-up class: those pixels fall into grassland or bare ground.
7
Climate by mid-century
In the 0.5° model cell that holds the point, the median of 9 models gives a mean temperature of −5.7 °C in the reference climate and −2.8 °C in 2041–2070 under SSP2-4.5 (+2.9 °C); under SSP5-8.5, −1.9 °C (+3.9 °C). Annual rainfall, 426 mm, changes by +6.3% under SSP2-4.5 and +5.3% under SSP5-8.5. Days above 35 °C: 0.0 a year now and 0.0 under SSP2-4.5. Days above the local 95th percentile of the fire weather index go from 17.5 to 24.8 a year under SSP2-4.5. In the Köppen-Geiger classification, the point's pixel is Dfc (Cold, no dry season, cold summer (boreal)) today and becomes Dsc (Cold, dry summer, cold summer) under SSP2-4.5 in 2041–2070. Relief over a 5 km circle runs from −6 m to 33 m. This is the climate of one model cell, not of the disc or of the area where the species lives, and these are scenarios, not forecasts.
8
What land cover cannot show
The literature we read places the species' limit in sea ice, not on land. A land-cover map does not show sea ice or the timing of its formation and retreat. A tundra that does not change on the map therefore does not mean the bears are doing well. We read no source on current abundance in Western Hudson Bay.
9
References, pictures and data
At the end of the page are the references and sources, numbered in the order they appear in the text, and the full credit of each picture. The species records in the disc are in the GBIF download doi:10.15468/dl.3rj6wn.
Full citation of each picture at the end of the page.
What this story does not say
The discs are not the species' range. Every number comes from discs of 10 and 50 km around a point chosen where GBIF holds pictures of the species; they are neither the species' range nor the boundary of a protected area, and they change with another outline.
Records measure effort. GBIF holds what someone saw and published: more records say where and when people were looking, not that there are more animals.
Canopy loss is not deforestation. Hansen counts any tree-cover loss, from any cause, and does not say why it happened.
Land-cover maps are models. Each map has its own legend and error; the world map does not separate pasture, and the common groups are this series' choice.
A cell's climate, not a place's. The climate numbers are the median of the models in one grid cell, under emission scenarios; they are not a forecast for the disc.
The other threatened species carry the category GBIF attaches to the name, not checked on the Red List.
Generated by the same program for every story in the series, from the text and references written for this species and the numbers Naturametrics and GBIF computed; texts written with Claude, by Anthropic. No number comes from the model.
Versão em português.
McGeachy, D., Lunn, N. J., Richardson, E. S., & Derocher, A. E. (2024). Sea ice influence on male polar bear survival in Hudson Bay. Arctic Science, 10(3), 483-498. https://doi.org/10.1139/as-2023-0045
Atwood, T. C., Marcot, B. G., Douglas, D. C., Amstrup, S. C., Rode, K. D., Durner, G. M., & Bromaghin, J. F. (2015). Evaluating and ranking threats to the long-term persistence of polar bears. Open-File Report. https://doi.org/10.3133/ofr20141254
GBIF.org (16 November 2026) GBIF Occurrence Download https://doi.org/10.15468/dl.3rj6wn. Download of the Ursus maritimus records in the 50 km disc, with each dataset's licence. The chart counts come from the GBIF search API, with the same filters, queried on 2026-11-16.
GBIF.org occurrence search with the iucnRedListCategory filter (CR, EN, VU) and the speciesKey facet, and the species API for the names, queried on 2026-11-16. The category is the one GBIF attaches to the species name in its backbone; it was not checked species by species on the IUCN Red List. No download DOI.
Hermosilla, T., Wulder, M. A., White, J. C., Coops, N. C., & Hobart, G. W. (2018). Disturbance-informed annual land cover classification maps of Canada's forested ecosystems for a 29-year Landsat time series. Canadian Journal of Remote Sensing, 44(1), 67-87. https://doi.org/10.1080/07038992.2018.1437719 Canada's forested-ecosystem land-cover series (NTEMS VLCE2, projects/sat-io/open-datasets/CA_FOREST_LC_VLCE2), read with the Naturametrics disc.
Hansen, M. C., Potapov, P. V., Moore, R., Hancher, M., Turubanova, S. A., Tyukavina, A., Thau, D., Stehman, S. V., Goetz, S. J., Loveland, T. R., Kommareddy, A., Egorov, A., Chini, L., Justice, C. O., & Townshend, J. R. G. (2013). High-resolution global maps of 21st-century forest cover change. Science, 342(6160), 850-853. https://doi.org/10.1126/science.1244693 Global Forest Change v1.13 (UMD/hansen/global_forest_change_2025_v1_13), loss 2001–2025, 2000 canopy of at least 30%.
Giglio, L., Justice, C., Boschetti, L., & Roy, D. (2021). MODIS/Terra+Aqua Burned Area Monthly L3 Global 500m SIN Grid V061 [Data set]. NASA Land Processes Distributed Active Archive Center. https://doi.org/10.5067/MODIS/MCD64A1.061 MCD64A1 monthly burned area, 500 m, summed by year.
Natural Resources Canada, NTEMS. Canada forest age (CANADA/NFIS/NTEMS/CA_FOREST_AGE in Earth Engine), age as of 2019; Naturametrics Canada page.
Zhang, X., Zhao, T., Xu, H., Liu, W., Wang, J., Chen, X., & Liu, L. (2024). GLC_FCS30D: the first global 30 m land-cover dynamics monitoring product with a fine classification system for the period from 1985 to 2022 generated using dense-time-series Landsat imagery and the continuous change-detection method. Earth System Science Data, 16(3), 1353-1381. https://doi.org/10.5194/essd-16-1353-2024 Annual world land-cover map GLC-FCS30D, 30 m (Earth Engine: projects/sat-io/open-datasets/GLC-FCS30D/annual), Naturametrics world page.
Thrasher, B., Wang, W., Michaelis, A., Melton, F., Lee, T., & Nemani, R. (2022). NASA Global Daily Downscaled Projections, CMIP6. Scientific Data, 9, 262. https://doi.org/10.1038/s41597-022-01393-4 NEX-GDDP-CMIP6, 0.5° cells; the fire weather index comes from NEX-GDDP-FWI (0.25°), read from muda-clima's world blocks.
Beck, H. E., McVicar, T. R., Vergopolan, N., Berg, A., Lutsko, N. J., Dufour, A., Zeng, Z., Jiang, X., van Dijk, A. I. J. M., & Miralles, D. G. (2023). High-resolution (1 km) Köppen-Geiger maps for 1901–2099 based on constrained CMIP6 projections. Scientific Data, 10, 724. https://doi.org/10.1038/s41597-023-02549-6 Köppen-Geiger class at the point's 1 km pixel.
European Space Agency, & Airbus (2022). Copernicus DEM [Data set]. https://doi.org/10.5270/ESA-c5d3d65 Copernicus GLO-30 (COPERNICUS/DEM/GLO30_2024_1), relief over a 5 km circle.
Pictures
Robert McLeod (2014). [Photograph of Ursus maritimus, Churchill (CA)]. iNaturalist research-grade observations. https://www.inaturalist.org/observations/170337747. Licence CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). Via GBIF.org, occurrence https://www.gbif.org/occurrence/4145246580.
Ann McKenzie (1990). [Photograph of Ursus maritimus, Churchill (CA)]. iNaturalist research-grade observations. https://www.inaturalist.org/observations/393638806. Licence CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). Via GBIF.org, occurrence https://www.gbif.org/occurrence/6498220737.
Laura Maskell (2026). [Photograph of Ursus maritimus, Churchill (CA)]. iNaturalist research-grade observations. https://www.inaturalist.org/observations/382357906. Licence CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Via GBIF.org, occurrence https://www.gbif.org/occurrence/6452190094.
Emile Chamberland (2026). [Photograph of Ursus maritimus, Churchill (CA)]. iNaturalist research-grade observations. https://www.inaturalist.org/observations/388790275. Licence CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Via GBIF.org, occurrence https://www.gbif.org/occurrence/6479149592.
Sean Frey (2024). [Photograph of Ursus maritimus, Churchill (CA)]. iNaturalist research-grade observations. https://www.inaturalist.org/observations/236774274. Licence CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Via GBIF.org, occurrence https://www.gbif.org/occurrence/4926045338.
Barry Stewart (2024). [Photograph of Ursus maritimus, Churchill (CA)]. iNaturalist research-grade observations. https://www.inaturalist.org/observations/362468329. Licence CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Via GBIF.org, occurrence https://www.gbif.org/occurrence/6320128412.