Room for Caribou

Woodland caribou · Ontario · 2015–2025

Room for Caribou

Woodland caribou need large areas of undisturbed forest. We measured how much is left in seven Ontario ranges, every year from 2015 to 2025.

Ranges
7Berens to Kesagami
Area
–
inside the managed forest
Years
112015 to 2025
Explore the ranges

Caribou: Alain Caron, Wikimedia Commons, CC BY-SA 4.0 · Earth: NASA Blue Marble

Research questions

Three questions, three answers

  1. Question 1How disturbed are the seven ranges, and how did that change from 2015 to 2025?Results →
  2. Question 2How much of the disturbance is anthropogenic, and how much is wildfire?Status in 2025 →
  3. Question 3How much do newer public records change the published assessment?Comparison →

Scientific basis

Why undisturbed forest matters

Boreal woodland caribou are listed as Threatened in Canada. They persist where large areas of mature forest remain undisturbed. Logging, roads and fire create young forest that supports moose and deer; these sustain more wolves, and caribou calves are lost to the added predation (Environment Canada 2011).

  1. HabitatMature conifer forest and treed wetland, where caribou find lichen and space away from predators. Mapped from Canada’s 2019 land cover, as in Mackey et al. (2024).
  2. 500 m buffer zoneCaribou avoid human features and predation risk is higher near them, so logging, roads, mines and settlements count with a 500 m buffer zone (Environment Canada 2011).
  3. FireBurns count without a buffer zone for up to 40 years while the forest regrows (Environment Canada 2011). Here, fires from 1985 on count in every year.
  4. 35% thresholdA range with more than 35% of its area disturbed has less than a 60% chance of supporting a self-sustaining population (Environment Canada 2011; ECCC 2020).
  5. Core habitatUndisturbed habitat, less a 30 m edge, in patches of 100 ha or more.

Full method →

Overview

Every range exceeds the 35% threshold

Each range in 2025 click a range or a ring for its numbers
2025
Anthropogenic disturbance Disturbed by wildfire Anthropogenic and wildfire Not disturbed Core habitat
What does “both” mean?A place counts as both when a fire and a human feature overlap: for example, a burn that lies within 500 m of a road or cutblock, or a stand logged after it burned. It is counted once, as anthropogenic.
Map: where each kind of disturbance is, and the core habitat left in blocks of 100 ha or more. Rings: share of the .

Canada considers a caribou range with more than 35% of its area disturbed unlikely to keep a self-sustaining herd. Logging, roads and wildfire open up the forest and bring in moose, wolves and bears.

Managed forest, lakes included is the published study’s basis and our default. Whole range is how Environment and Climate Change Canada and Ontario report. Land only leaves out open water, where caribou do not live. See how our numbers compare.

Built from 15 public datasets

All datasets and years →

Results, 2015–2025

Disturbance grew, and most of it is anthropogenic

How we checked the method →

Share of each range disturbed, 2015–2025 % of · dashed line = 35% threshold

Anthropogenic or wildfire?

The rings and range card at the top show this split for any year.

How we checked the new logging records →

Caribou range

Berens

See the change from 2020 to 2025

How the records after 2020 were tested →

drag the slider
2020 2025

Discussion

What this means for caribou

    Keep in mind

    • Percentages refer to each range within Ontario’s managed forest, lakes included, as in the published study.
    • Logging records end in 2023, so 2024 and 2025 show little new logging. Fire is mapped to 2025.
    • Roads are treated as permanent disturbance, including closed roads.

    What comes next

    • Add planned harvest from forest management plans to project future disturbance.
    • Update the maps each year as new harvest reports and fire maps are released.

    How our numbers compare

    Each source measures a different area. Switch between the managed forest, as in the published study, and the whole range, as in the federal and Ontario reports. Striped bars are figures we measured from that source's own maps or rebuilt with its rules.

    Show the numbers
    Why the numbers differ
    • Area. The whole range includes undisturbed northern land outside the managed forest, which lowers the share.
    • Roads. ECCC's 2020 lines hold far fewer forest roads than Ontario's MNRF road layer, and every road adds a 500 m buffer zone.
    • Logging. ECCC's cutblocks cover more ground than Ontario's records, because their outlines are broader (compare them in the Data tab).
    • Same rules. All sources use a 500 m buffer zone around human disturbance, none around fire, and the same 35% threshold.
    • Class 18. Ontario's land cover has a "disturbance" class that mixes cuts and burns. Inside a fire mapped from 1985 on, this study counts it as that fire, without a buffer zone. This lowers the share slightly, most in Berens.
    • 2025. No ECCC map for 2025 was available, so 2025 is shown for this study only.
    • Environment Canada (2011). Scientific Assessment to Inform the Identification of Critical Habitat for Woodland Caribou (Rangifer tarandus caribou), Boreal Population, in Canada: 2011 Update. Ottawa. PDF
    • Environment and Climate Change Canada (2020). Amended Recovery Strategy for the Woodland Caribou (Rangifer tarandus caribou), Boreal Population, in Canada. Species at Risk Act Recovery Strategy Series. canada.ca. ECCC figures are computed from its 2020 disturbance layers.
    • Fryxell, J.M. et al. (2020). Anthropogenic disturbance and population viability of woodland caribou in Ontario. Journal of Wildlife Management. doi:10.1002/jwmg.21829
    • Mackey, B. et al. (2024). Assessing the cumulative impacts of forest management on forest age structure development and woodland caribou habitat in boreal landscapes: a case study from two Canadian provinces. Land 13, 6.
    • Ontario Ministry of Natural Resources and Forestry (2014). Integrated Range Assessments for Woodland Caribou and their Habitat, e.g. Kesagami Range 2010; summary in the State of the Woodland Caribou Resource Report.

    References

    Sources cited

    • Environment Canada (2011). Scientific Assessment to Inform the Identification of Critical Habitat for Woodland Caribou (Rangifer tarandus caribou), Boreal Population, in Canada: 2011 Update. Ottawa. PDF
    • Environment and Climate Change Canada (2020). Amended Recovery Strategy for the Woodland Caribou (Rangifer tarandus caribou), Boreal Population, in Canada. Species at Risk Act Recovery Strategy Series. canada.ca. ECCC figures are computed from its 2020 disturbance layers.
    • Fryxell, J.M. et al. (2020). Anthropogenic disturbance and population viability of woodland caribou in Ontario. Journal of Wildlife Management. doi:10.1002/jwmg.21829
    • Mackey, B. et al. (2024). Assessing the cumulative impacts of forest management on forest age structure development and woodland caribou habitat in boreal landscapes: a case study from two Canadian provinces. Land 13, 6.
    • Ontario Ministry of Natural Resources and Forestry (2014). Integrated Range Assessments for Woodland Caribou and their Habitat, e.g. Kesagami Range 2010; summary in the State of the Woodland Caribou Resource Report.

    Data sources and their providers are listed under Datasets.

    ← Back to the findings

    DataEvery dataset, and how we checked itThe years each source covers, whether the datasets agree, a satellite check of the new records, and every dataset by name. → ApproachHow the maps are madeThe processing chain, one window followed step by step, and the published result rebuilt from the same data. → ResearchQuestions, scientific basis and referencesThe three research questions with their answers, why undisturbed forest matters for caribou, and the sources cited. →

    Method and workflow

    How the maps are made, and how we repeated the study

    We use the method of Mackey et al. (2024). First we rebuilt their 2020 map from their data. Then we added newer logging (ARI) and fire (NBAC) records. The Data tab shows how we checked them.

    Each circle is one type of data. The label on each line is the rule we use. Point at a circle to learn more.

    DATA RULES YEARLY MAP RESULTS
    goes into the map used only to check published study data the published study used added added to reach 2025 check used only to check
    Close-up

    Land cover, 2019

    Follow one 27 × 27 km window in the Churchill range. Every 30 m pixel carries a class from VLCE2, NRCan's national forest land-cover map.

    Potential habitat

    Select potential habitat

    Only conifer forest and treed wetland count as potential habitat. Other land-cover classes are excluded before disturbance is assessed.

    Potential habitat: – of the window

    Disturbance, 2020

    Combine all disturbance layers

    A pixel is disturbed if any dataset maps disturbance there: logging since 1980 from FRI or NTEMS harvest, planted stands, roads, mines, towns, farms, corridors and OLCC's disturbance class, each with a 500 m buffer zone, plus wildfire. Overlaps count once.

    Disturbed: – of the window

    Undisturbed habitat

    Remove disturbed habitat

    What remains is potential habitat that no disturbance reaches. In this window, – of the potential habitat is lost.

    Undisturbed habitat: – of the window

    Core habitat

    Identify core habitat

    A 30 m edge is removed and adjacent pixels are grouped into patches. Patches under 100 ha are excluded; the rest are grouped at 500, 1,000 and 10,000 ha.

    Core habitat: – of the window · – of it in patches over 1,000 ha

    Land cover, 2019 Churchill range · 27 × 27 km · 30 m pixels
    Land cover in the close-up window: mostly conifer and mixedwood forest with lakes. Potential habitat: conifer forest and treed wetland in green. Disturbance in 2020 shown over potential habitat. Undisturbed habitat in green; habitat lost to disturbance in pale colour. Core habitat coloured by patch size, with eroded edges in pale green.

    The reproduction

    First, we repeated the published study with its own data and rules

      Technical details

      Reproduction. Same inputs as the study: 17 Forest Resources Inventory packages, the national Landsat harvest layer 1985–2020, MNRF and national road networks, the Ontario Land Cover Compilation (classes 18, 25, 27, 28), and the National Fire Database 1985–2018. We checked our map pixel by pixel against the layers the authors shared. The remaining 1 to 3 points come mostly from roads, which are only published as current editions.

      Overlap. On the 30 m grid inside each range's managed forest, every dataset's mapped area is split into the part also mapped by all the other datasets of the same type, by some of them, or by none. Another dataset counts as mapping a pixel when it maps something within 60 m, so that polygon edges, Landsat pixels and road centre lines drawn slightly apart still match. Comparisons that include ECCC 2020 are limited to the federal range, where ECCC maps.

      Rules for the 2015–2025 maps. Fire counts from 1985 in every year (NFDB and NBAC). ARI logging from 2013 on is used only where at least half of a stand shows a clear drop in the forest signal on Landsat or Sentinel-2; 2,609 stands (40,946 ha) without a visible cut are excluded. OLCC's disturbance class (18) mixes cuts and burns, so where NFDB or NBAC maps a fire from 1985 on it is counted as fire (no 500 m buffer zone); everywhere else it stays anthropogenic, as in the published study.

      Satellite check. Random spots inside ARI records: about 400 per record type on Landsat (logging only where FRI and NTEMS have none) and 190 to 350 per type on Sentinel-2, with cloud-free summer images from two years before to three years after the recorded year (Landsat 8/9 for 2013–2020, Sentinel-2 for 2018–2024). A spot shows forest loss if its Normalized Burn Ratio falls by more than 0.1 at some point in that six-year window, so cuts recorded a year or two early or late still count. Logging recorded by both FRI and NTEMS harvest is checked the same way as a reference: 94% of it shows loss, and the rest is mostly partial cuts, cuts smaller than a pixel, or stands already open. The full check then tested every ARI logging stand from 2013 on over its whole area, with Landsat and, from 2018, Sentinel-2. Blowdown and drought records are excluded because Canada's caribou method counts only human disturbance and fire.

      Code and notebooks: github.com/RSL-SLU/caribou

      ← Back to the findings

      Data sources

      Public records, from 1980 to 2025

      We start from the data a published study used, Mackey et al. (2024), and add two datasets that carry it to 2025: ARI for logging and NBAC for fire. Satellite images are used only to check the records. Below: which years each source covers, two checks of the data, and every dataset by name.

      Which years each source covers hover a bar
      Used by the published study Added to reach 2025 Used only to check

      Check 1 · Agreement between datasets

      Datasets that map the same disturbance agree

      For each disturbance type, datasets were compared pixel by pixel on the 30 m grid, with a 60 m tolerance. Each card gives the share of one dataset that the others also map.

        Agreement by range

        Switch datasets on or off. Places mapped by two or more are green.

        hover the map
        Sentinel-2 · summer 2024 (gaps from 2023) · false colour
        Notes

        used goes into the habitat maps · check only is compared with them but not counted.

        A high “mostly unique” share can reflect a more complete dataset, such as MNRF roads with its forest roads.

        Check 2 · are the new records real?

        Checking ARI against Landsat and Sentinel-2 images

        1. Purpose. ARI is the only logging record after 2020, and part of it is in no other dataset (Check 1). Each recorded cut should be visible from space.
        2. Sample. Random points inside ARI records, per record type: about 400 with Landsat (2013–2020; logging absent from FRI and NTEMS only) and 190–350 with Sentinel-2 (2018–2024).
        3. Imagery. Cloud-free summer composites from two years before to three years after the year of record.
        4. Measure. Normalized Burn Ratio (NBR): high over closed forest, it falls sharply when trees are cut or burned.
        5. Rule. A record shows loss if NBR falls by more than 0.1 between any two of these summers, so a cut dated a year early or late still counts.
        6. Reference. Logging recorded by both FRI and NTEMS harvest is tested the same way (grey bar).
        7. Full test. All 16,369 ARI logging stands from 2013 on were tested over their whole area. A stand is retained if at least half of its pixels show loss: 13,760 are retained.

        NBAC fire is not tested here: it is itself mapped from satellite images, so the test would confirm it by construction. Check 1 shows it matches the NFDB and ARI fire records instead.

        Point test on 1,990 sampled records: share showing forest loss
        Grey: reference logging recorded by both FRI and NTEMS harvest Hover a bar for the details.

        Satellite evidence for each record

        Sampled records: ARI records drawn at random for a point test, shown by range and record type; filled where loss is detected. All ARI logging stands: every logging stand recorded from 2013 to 2023, filled where the full-stand test retains it. Hover a shape for its forest signal; double-click to play its summers.

        Sentinel-2 · summer 2024 (gaps from 2023) · false colour

        Datasets

        Every dataset, by name

        The short name in the first column is the one used throughout this study.

        NameDataset and providerYearsFormatUsed forRole
        RangesWoodland caribou range boundaries
        Land Information Ontario        		currentpolygonsThe seven ranges analysedPublished study
        FMUForest Management Units
        Land Information Ontario        		2025polygonsThe managed forest; every percentage is of the range inside itPublished study
        VLCE2Canada forest land cover 2019 (Virtual Land Cover Engine)
        Natural Resources Canada        		201930 m rasterPotential caribou habitat: conifer forest and treed wetlandPublished study
        FRIForest Resources Inventory, Packaged Products v2 (17 management units)
        Ontario MNRF        		logging 1980 to each inventory yearpolygonsLogging and planted stands; unclassified corridorsPublished study
        NTEMS harvestLandsat-derived forest harvest 1985–2020 (Hermosilla et al.)
        Natural Resources Canada        		1985–202030 m rasterLogging; its year is used where FRI and NTEMS harvest both record loggingPublished study
        MNRF roadsMNRF Road Segments
        Land Information Ontario        		by year built, to 2025linesRoads, 500 m buffer zonePublished study
        NRNNational Road Network v18, Ontario
        Statistics Canada        		currentlinesRoads, 500 m buffer zonePublished study
        OLCCOntario Land Cover Compilation v2.0 (classes 18, 25, 27, 28)
        Ontario MNRF        		imagery 2000–201115 m rasterMines, towns, farms and land-cover disturbance, 500 m buffer zone. Its disturbance class (18) mixes cuts and burns: inside a fire from 1985 on (NFDB or NBAC) it counts as firePublished study
        NFDBNational Fire Database fire polygons
        Natural Resources Canada (CWFIS)        		1972–2024polygonsFire from 1985 (1985–2018 for the reproduction)Published study
        ARIAnalysis Ready Inventory 2025
        Ontario MNRF        		annual reports to 2023polygonsLogging the other sources miss, and all logging after 2020 (from 2013 on, only stands seen from space); its fire and blowdown records are only comparedAdded to reach 2025
        NBACNational Burned Area Composite
        Natural Resources Canada (CWFIS)        		1980–2025polygonsFire from 1985, including 2025; mapped from satellite imagesAdded to reach 2025
        NTEMS harvest 2022Landsat-derived forest harvest 1985–2022
        Natural Resources Canada        		1985–202230 m rasterChecking ARI after 2020 and its yearCheck only
        ECCC 2020Anthropogenic disturbance footprint 2020, from 30 m Landsat-8
        Environment and Climate Change Canada        		2020lines and polygonsChecking logging, roads, settlements and mines; compared with our resultsCheck only
        LandsatCollection 2 Level-2 surface reflectance
        NASA / USGS        		2011–2025 images30 m imagesChecking ARI logging from 2013 onCheck only
        Sentinel-2Level-2A surface reflectance
        ESA Copernicus        		2016–2025 images10–20 m imagesChecking ARI logging from 2018 on; satellite base map (summer 2024)Check only

        ← Back to the findings