Beaver dam capacity of Canada’s boreal plain in response to environmental change


Environmental changes are altering the water cycle of Canada’s boreal plain. Beaver dams are well known for increasing water storage and slowing flow through stream networks. For these reasons beavers are increasingly being included in climate change adaptation strategies. But, little work focuses on how environmental changes will affect dam building capacity along stream networks. Here we estimate the capacity of the stream network in Riding Mountain National Park, Manitoba, Canada to support beaver dams under changing environmental conditions using a modelling approach. We show that at capacity, the park’s stream network can support 24,690 beaver dams and hold between 8.2 and 12.8 million m3 of water in beaver ponds. Between 1991 and 2016 the park’s vegetation composition shifted to less preferred beaver forage, which led to a 13% decrease in maximum dam capacity. We also found that dam capacity is sensitive to the size of regularly-occurring floods—doubling the 2-year flood reduces the park’s dam capacity by 21%. The results show that the potential for beaver to offset some expected climatic-induced changes to the boreal water cycle is more complex than previously thought, as there is a feedback wherein dam capacity can be reduced by changing environmental conditions.


Canada’s boreal forest has long been important beaver, Castor canadensis, habitat1. Intensive trapping of beaver in the 1700s through the 1800s caused near extirpation of the species in the boreal forest2. Population recovery began in the 1930s via re-introduction and conservation programs3. Beaver are keystone in supporting food security of Indigenous people residing in the boreal forest through providing a source of meat and being ecologically influential4,5. They are viewed as an ecologically influential species as they profoundly alter the aquatic ecosystems they occupy2 with benefits for freshwater biodiversity6 and terrestrial wildlife7. Beaver alter aquatic ecosystems primarily through engineering activities—dam and canal building—which modifies ecosystem-forming processes8. For example, beaver dams raise and stabilize water tables9,10, alter stream hydrographs11,12, enhance channel and riparian area sediment retention13,14 and create hydrologically complex, multi-channel networks9,15. The suite of changes beaver dams make to aquatic ecosystems creates a multitude of desired ecological feedbacks that collectively enhance ecosystem resilience to disturbance16.

Environmental changes resulting from the cumulative impacts of disturbance-recovery dynamics and climate change17 are expected to considerably alter the ecological and hydrological character of the boreal forest over time18. Beaver are likely an important part of enhancing the resiliency of the boreal forest to environmental change19,20. Yet, recent and continued environmental changes will also impact the beaver population. Increases in the occurrence and severity of wildfires, for example, are expected in the future which will change the vegetation composition of boreal forests21. Fires renew beaver food sources, especially willow and aspen which they prefer, providing quality habitat for beaver 5–30 years afterwards22,23,24,25. Further, the growing oil and gas extraction industry in the western portion of Canada’s boreal forest26 is causing loss of beaver habitat but creation of new habitat in engineered landforms27. The way in which the beaver population will respond to the set of ongoing environmental changes is likely complex. For example, beavers may respond to climate change by modestly expanding their range, especially northward28, and densifying in their current range29.

Many studies have compared beaver site occupancy with habitat characteristics; recent examples include Smeraldo et al.30 and Scrafford et al.31. But, few studies have asked how changing beaver habitat characteristics owing to environmental change will affect how many and where beaver dams are built along stream networks, information critical to understanding the effects of climate change on water resources. In this study we assess the landscape capacity of the boreal forest to support beaver dams under changing environmental conditions. We use the example of Riding Mountain National Park in Canada (Fig. 1) because the park is located at the southeastern limit of the boreal plain which is most vulnerable to climate change18; the park also has long-term records of beaver caches. Beaver were extirpated from the park and adjacent areas via trapping before 1930 and reintroduced in 194732. Ongoing aerial surveys of food caches since 1973 are used by the park to track beaver population dynamics33. To achieve our aim, we ran the BRAT model34 (see “Methods”) across Riding Mountain National Park under various scenarios to assess: (1) the capacity of the park’s 2604 km of streams to support beaver dams; (2) how changes in forest composition over the past three decades have affected beaver dam capacity; and (3) how climatic-induced changes in streamflow will influence beaver dam capacity. We find that the park can support a large number of beaver dams, and that beaver ponds store a considerable volume of water. However, changes to park vegetation cover and the size of regularly occurring floods will reduce the number of beaver dams the park can support. The approach used should hold significant value for use elsewhere, not only for understanding changes in beaver habitat and dam capacity under previously recorded change, but also for estimating changes under future land use and land cover change scenarios.

Figure 1

The study site, Riding Mountain National Park. Depicted in the park is the stream drainage network occurring within each of the four main physiographies (main image). The inset situates the park (black square) within Canada’s Boreal Plains ecozone (shaded).


Beaver dam capacity and pond water storage

The modelling results indicate Riding Mountain National Park can support a large number of beaver dams. We estimate the park has a dam capacity of 24,690 dams or 9.5 dams per stream km based on 2016 imagery (Fig. 2). At 70% of the stream network (1833 km), the frequent dam density category is the largest. Little of the stream network (3%) was in the rare and none dam density categories. Using the park modelled dam capacity estimate, a bootstrapped median beaver pond size of 1419 m2 (95% confidence interval, CI, of [1157, 1704 m2]) and the water storage algorithm of Karran et al.35 (see “Methods”), we estimate the park has a beaver pond storage capacity 10.4 million m3 of water (CI of [8.2, 12.8 million m3]).

Figure 2

Modelled beaver dam capacity (by capacity category) at the reach scale (300 m segment) for the stream network within Riding Mountain National Park. Maximum beaver dam density by physiography is reported.

Modelled dam density varies across the park. Dam capacity is greatest in the plateau (13,105 dams) and lowest in lowlands (2389). The main cause of this difference is the length of the stream network. When dam capacity is scaled to stream length, there is less difference among physiographies. The upper limit of dam density is highest (12.0 dams per stream km) in the hummocky physiography and lowest in the plateau and escarpment physiographies (~ 9.2 dams per stream km in each).

Beaver dams, determined from observations made in survey blocks scattered across the park (120 km2; see “Methods”), are present at approximately 22% of capacity (Fig. 3). There was, however, high variation in dam capacity by physiography. Beaver dams in the escarpment occur at the lowest percentage of capacity (10%) and in the hummocky at the highest percentage of capacity (51%). Dams occur in the lowland and plateau at 30% and 32% of capacity, respectively. One survey block (no. 24) had more beaver dams than the modelled capacity (150%)—this survey block was in the hummocky physiography. The BRAT model used to estimate beaver dam capacity only considers dams occurring along the river network. However, in conducting the analysis of beaver dam counts in the survey blocks, we observed a number of beaver dams in off-channel locations (Fig. 4), especially in the hummocky physiography and along the hummocky-plateau boundary.

Figure 3

Beaver dams, determined from observations made in the long-term survey blocks across the park (see S1), are present at approximately 22% of capacity based on 2004 high-resolution imagery the 2003 BRAT results. Average proportion of dams in relation to modelled capacity for each physiography is: 10% escarpment, 51% hummocky, 30% lowlands, and 32% plateau.

Figure 4

Aerial photograph (0.63 m resolution) from 2004 showing examples of beaver dams (circled) occurring in off-channel locations. The air photo is owned by Riding Mountain National Park and is used with permission. The blue line depicts the stream channel, the shapefile for which was sourced from the publicly available Manitoba Land Initiatives archive, The map was generated using ESRI ArcMap 10.5.

Impacts of changing vegetation composition on beaver dam capacity

Vegetation cover of Riding Mountain National Park has changed over the last 25 years (Fig. 5). Overall changes were small—there was a 2.8% change in vegetation cover from 1991 to 2003, and an additional 3.4% change from 2003 to 2016. Much of the change between 1991 and 2003 occurred in the western part of the park in the hummocky and plateau physiographies while changes from 2003 to 2016 primarily occurred in the central and southeastern areas of the park in the plateau physiography. The largest vegetation change from 1991 to 2003 was a 23.2 km2 increase in marsh. The major vegetation change from 2003 to 2016 was a 28.2 km2 decrease in mixedwood. In places where vegetation cover change occurred, preferred beaver forage was reduced and replaced with less suitable beaver forage (Fig. 6).

Figure 5

Places where major land cover change was detected in Riding Mountain National Park between 1991 and 2003 (a), and 2003 and 2016 (b).

Figure 6

Suitability of vegetation cover for beaver foraging in places identified as changed between 1991 and 2016 on Fig. 5.

Beaver can only build dams in habitats where there is suitable and sustainable forage34,36 and so we estimated beaver dam capacity for the park by running the BRAT model with the 1991, 2003 and 2016 vegetation composition. The model results indicate streams in Riding Mountain National Park could support 14% more beaver dams in 1991 than in 2003 or 2016 (Fig. 7). The largest changes from 1991 and 2003 occurred within the pervasive and occasional dam density categories. Specifically, there was three times greater stream length in the pervasive dam density category in 1991 than in 2003 or 2016, which occurred primarily in the southern part of the park. Further, there was a third less stream length in the occasional dam density category in 1991 than in 2003 or 2016. Correspondingly, the estimated upper limit of water storage capacity in beaver ponds lowered from 12.0 million m3 in 1991 to 10.4 million m3 in 2003, a reduction of 13.3%. Beaver dam capacity and beaver pond water storage capacity in the park were similar when the model was run with 2003 vs 2016 vegetation cover (median ~ 10.4 million m3 of water for each year).

Figure 7

Comparison of beaver dam capacity by physiography as modelled using the vegetation cover in 1991, 2003 and 2016.

Impacts of stream flow changes on beaver dam capacity

In addition to suitable forage, beaver require a reliable water source in order to build dams. On streams, what is suitable is perennial flow without large floods happening too often such that dams regularly fail37. The BRAT model uses the stream power associated with the two-year recurrence interval peak flood (Q2) to infer the likelihood that a beaver dam will persist once built34. Knowing that climate changes are already leading to wetter conditions in this part of Canada38 and are likely to continue into the future19, we ran scenarios in the BRAT model across the park wherein only Q2 was modified to assess how climatic-induced changes to streamflow might influence beaver dam capacity (see “Methods”). We found that smaller increases in the magnitude of Q2 modestly reduces the capacity of the park’s stream network to support beaver dams (Fig. 8). Increasing Q2 by 100% though reduces beaver dam capacity in the park by 21%. The decreases in dam capacity were most noticeable in the frequent dam density category, which decreased from comprising 70% of the stream network to 52% of it with a doubling of Q2. The stream segments originally classified in the frequent dam density category under Q2 shifted to being in the occasional (increase of 13%) and rare dam density categories (increase of 8%) in the 100% Q2 scenario.

Figure 8

Model estimates of number of beaver dams in each dam density category for the stream flow change scenario analysis.


The model predictions of beaver dam capacity presented here have at least three important implications. First, Riding Mountain National Park can support a large number of beaver dams, although dam capacity varies across the park. The many ecological and hydrological benefits provided by beaver dams are closely related to their frequency along a stream course39. While high, an average of nine dams per stream km density is not unrealistic for this region. Historic (pre fur trade) beaver occupation was at least this high40, and there are reports of current dam density exceeding 10 per km of stream in the literature11,41. High dam density is possible in Riding Mountain National Park, like much of the boreal forest in Canada, because it has an abundance of beaver-preferred vegetation. Beaver prefer aspen, poplar and cottonwood trees and willow, using these species for food and construction materials42,43,44. Mixedwood stands—comprised of aspen and poplar and an understory of willow and beaked hazelnut45—are extensive in the park. If the park reached its dam capacity, we estimate beaver dams would impound between 8.2 and 12.8 million m3 of water. To put that in perspective, the province of Manitoba is well known for hydroelectric dams. One of the largest hydroelectric dams in the province is the Limestone; it has a water storage capacity of just under 3 million m346. Thus there is potential for impoundment of large volumes of water in beaver ponds with considerable impact to the park’s streamflow regime. Beaver-mediated enhanced water storage in the boreal plain is likely to become increasingly important with climate change as shallow ponds dry out20,47.

The distribution of beaver dams across the park is spatially variable. Our study area has different physiographies that differ in their topography and geology. Although all four physiographies have suitable beaver habitat, model runs indicate maximum beaver dam density in the escarpment and plateau physiographies is lower than in the hummocky and lowland physiographies. The differences in dam density results from the combined influence of availability of preferred vegetation adjacent to the stream network composition and the steepness of stream channels22. Where beaver-preferred vegetation is widely available near streams, flatter landscapes tend to support higher dam densities48. The hummocky and lowland physiographies are both low gradient environments. There is growing concern from farmers, particularly those with land to the north of the park, that having too many beaver dams increases the risk of flooding on their fields. This concern is not unfounded, as we show dam capacity is high along the northern boundary of the park as this is where the lowlands physiography occurs. Given there is the potential for a large volume of water to be impounded near the northern park boundary, farmers are pressuring park managers to remove beaver dams from this area as they fear the kinds of destructive outburst floods created by beaver dam failure that Hillman49 described.

Current beaver dam density on streams in Riding Mountain National Park is only 22% of the capacity predicted by the BRAT model. It seems unlikely that the beaver population is still recovering from the fur trade as beaver cache counts in the park are relatively steady50. There are several other possible reasons why beaver build fewer dams than the model predicts can be build. Every model has limitations. Some of the parameters the BRAT model does not account for are wet and dry cycles, common in the boreal plain18, or the fine-scale variations in habitat features, like the presence of deep pools51, that are important in influencing the number of dams a beaver colony will build along a stream reach at any time. Some of these challenges may be overcome by calculating the probability of dam construction within a Bayesian framework, such as was recently carried out in Graham et al.52. Also, beaver in Riding Mountain National Park compete with ungulates (primarily moose and elk) for woody resources, which would limit their population and thus ability to build dams53. Further, the park is actively managing beaver in places near hiking trails, buildings and roads. While management practices include both killing beaver and dismantling dams (R. Baird, pers. comm.), these practices are occurring in localized areas only and so are unlikely to significantly impact the overall number of beaver dams in the park.

Importantly, we observed many beaver dams not on stream channels, especially in the hummocky physiography. These dams are occurring away from streams, similar to what is reported to occur in mountain environments54. This is not surprising as the boreal forest is known to consist of many drainage pathways that are not mapped as streams9. In particular, hummocky parts of the boreal landscape tend to have a disjointed drainage pattern and high density of open water bodies55. Beaver also readily inhabit ponds20 and groundwater seeps56 which are abundant in the boreal plain. Dams built in these environments are critical locations in boreal plains watersheds for surface water storage39, and they are also likely important places of groundwater recharge10. But, beaver dams not located within the stream network are not captured by the BRAT model. While these dams clearly have a different hydrological role in the landscape than those located on stream channels, they do contribute overall to beaver-mediated hydrological modification of the boreal forest and so improvements to the BRAT model are needed to make it a widely applicable tool.

The second implication of our results involves changes in forest composition that alter the park’s beaver dam capacity. Although we did not identify specific causes of forest change over the study period, the composition and structure of the boreal forest naturally changes over time. Boreal forests are disturbance-driven ecosystems with fire, for example, being an integral part of a healthy boreal ecosystem57. Riding Mountain National Park uses prescribed burning and also selective logging as part of their park management plan58. Fires can renew beaver food resources. However, aspen does not become abundant in the Canadian boreal forest until 10–15 years post fire disturbance59, and it might take several decades for beaver colonies to flourish post fire disturbance as population recovery lags vegetation recovery24. As well, beaver play a role in altering forest composition—they are well known to create and maintain vegetation conditions conducive to their reoccupation through flooding and selective foraging60. When beavers are trapped out of ecosystems, forests transition to conifer-dominated, which is not palatable to beaver61. Understanding how past forest change impacts beaver dam capacity over time is a crucial first step to understanding how beaver dam capacity could be impacted in the future, and is an area requiring further research.

The third implication of our work is that future changes in climate will impact the dam-building capacity of boreal beavers. Looking forward, the climate future of the boreal plains is expected to be quite different than it has been. A future warmer climate will alter the water cycle, including the precipitation to evaporation regime18,23,25,62. Already, this part of Canada’s boreal forest is noticeably wetter than it was a few decades ago39, which is causing nonlinear increases in streamflow63. The modelling results under scenarios of increased streamflow indicate we should expect a reduction in beaver dam capacity and thus in beaver-mediated water storage. The modelling results indicate that a doubling of the size of the 2-year flood will reduce beaver dam capacity in RMNP by 21%. The BRAT model classifies a stream reach as not capable of supporting beaver dams where stream power exceeds 2000 W/m; a threshold above which McFarlane et al.34 indicate beaver dams commonly fail. However, there is significant uncertainty around dam failure threshold. Some studies report that higher streamflow, especially during spring snowmelt, can lead to higher rates of beaver dam failure64 while other studies report low rates of dam failure during large floods65,66. Thus, the dam failure threshold probably varies widely based on the state of repair of a dam, the materials from which the dam was built and pond fullness at the onset of a rain event11.

Overall, the study results bring new insights into the ability of beaver to offset some of the predicted effects of climate change. Specifically, it is the dam modification of stream hydrology that has led to a growing reliance on beaver in climate change adaptation (and stream restoration) strategies21. However, there are numerous interacting processes regulating where and how many dams beaver built along stream networks51. Our model results suggest climatic changes to streamflow regimes and forest composition will cause a feedback whereby the capacity of stream networks to support beaver dams is diminished. Thus, our findings are inconsistent with the beaver population model predictions of Jarema et al.29 whereby climate changes will lead to substantial increases in beaver density within the interior of their range. However, a direct link between the beaver population and the number of dam built remains to be shown.


Study area

The study was carried out in Riding Mountain National Park, Canada, located at the southeastern limit of the boreal plain (Fig. 1). Riding Mountain National Park is 2976 km2 and is part of a larger biosphere reserve which encompasses the surrounding multi-use landscape. Riding Mountain National park has 2604 km of streams, distributed unevenly across its four physiographies: escarpment (721 km), lowlands (229 km), hummocky (208), and upland plateau (1446 km). The upland plateau comprises most of the park and is characterized by rolling hills with an uneven distribution of glacial till and thus consists of many poorly-drained wetlands67. The Manitoba escarpment is characterized by eroded shale bedrock dissected by deep ravines produced by streams draining the upland plateau. The lowlands district is located at the base of the escarpment and is dominated by eastern deciduous forests where flooding is common and fire disturbance is rare. The hummocky district occurs in the southern and western reaches of Riding Mountain National Park and is characterized by numerous wetlands created by melting blocks of ice left by the retreating glaciers. Forest cover in Riding Mountain National Park is boreal mixedwood. Key tree species include white spruce (Picea glauca), trembling aspen (Populus tremuloides), balsam poplar (P. balsamifera), paper birch (Betula papyrifera) and black spruce (Picea mariana)68. Dominant shrubs include beaked hazelnut (Corylus cornuta) and mountain maple (Acer spicatum) in well-drained sites, and alder (Alnus spp.) in moist sites58.

Modelling beaver dam capacity

We modelled beaver dam capacity using the Beaver Restoration Assessment Tool (BRAT) model34, version PyBRAT 3.0.17, available at: BRAT is a GIS-based model that combines stream segment slope, beaver vegetation preferences and streamflow information to predict the upper limits of stream reaches to support beaver dams, and then projects these limits onto a stream network. The BRAT model was chosen over other beaver occupancy models as it relies on datasets that are less precise than those required for statistically-based models, and so can be used across large areas and in landscapes with different characteristics. The requisite inputs to the beaver dam capacity model include a drainage network layer, vegetation type raster data of historic and existing conditions, a digital elevation model (DEM) raster, and streamflow (baseflow and peak flow) information throughout the drainage network. Methods by Macfarlane et al.34 were followed using ESRI ArcMap 10.5 with minor customizations including the preprocessing of BRAT vegetation and stream power data to customize the tool to the boreal forest generally and Riding Mountain National Park specifically.

Vegetation input

BRAT modelled output of dam capacity is driven by vegetation data and constrained by stream power and stream gradient. Historic and existing vegetation layers for a study site are required inputs to the BRAT model. BRAT was created to use the publicly available datasets in the United States. These are not necessarily available in Canada. So, we used previously classified landcover maps of Riding Mountain National Park from 1991, 2003, and 2016 provided by Riding Mountain National Park. The imagery analysis for 1991 was created using a 30 m resolution LANDSAT 5 image acquired in August of 1991. The vegetation layer from 2003 was created using LANDSAT 5 Thematic Mapper (TM) imagery at a 30 m resolution; the month of that imagery was acquired was not available. Imagery from 2016 was derived from Sentinel 2 satellite imagery from June and July 2016 at a 10 m resolution. Since the vegetation raster for 2016 was at a 10 m resolution, the aggregate tool in ESRI ArcMap 10.5 was used with a cell factor of ‘3’ and the aggregation type kept as the default (‘SUM’). The output was a reduced-resolution version of 30 m.

To tailor beaver vegetation selection based on vegetation common in the boreal forest, we modified Macfarlane et al.’s34 vegetation suitability classification. We added a “VEG_CODE” field in each of the three vegetation raster files by adding a field code to each layer’s attribute table within ESRI ArcMap. The ‘VEG_CODE” field allows for each landcover type to be customized to the beaver preference. A value of 0–4 was assigned to each vegetation classification type based on its suitability as dam building material, with 0 being ‘unsuitable’, 1 being ‘barely suitable’, 2 being ‘moderately suitable’, 3 being ‘suitable material and 4 being ‘preferred material’.

Stream power input

The BRAT model uses stream power to determine the likelihood of a beaver dam persisting along a stream reach during low flow (Qlow) and 2-year flood (Q2) conditions34. There are no stream gauges within the park. Thus, in order to apply the BRAT model in RMNP, we chose the nearest Water Survey of Canada hydrometric station (gauge 05LJ045) located to the north of the park and used the daily and annual discharge record from it (1979–2015) in a regional flood frequency analysis. To calculate Qlow, 7Q10 was calculated using the R package ‘Flowscreen’ which used the lowest discharge reading and the subsequent measurement for 6 days for each hydrologic year and then calculated the mean discharge. Using the Qlow and Q2 discharge values for gauge 05LJ045, we edited the ‘iHyd’ python script which is a part of the BRAT toolbox in ESRI ArcMap 10.5 in IDLE. The ‘iHyd’ tool prepares the hydrologic inputs to BRAT by taking the equations created for Qlow and Q2 and calculates the values using the drainage-discharge relationship. The tool adds on two additional fields in the attribute table called ‘iHyd_SPLow’ and ‘iHyd_SP2’. Stream power was subsequently calculated as:

$$\Omega = \rho \cdot {\text{g}} \cdot {\text{Q}} \cdot {\text{S}}$$

where Ω is the total stream power (W/m), ρ is water density (1000 kg/m3), g is gravitational acceleration (9.8 m/s2), Q is discharge (m3/s) and S is steam bed slope, which was calculated using the DEM to estimate the highest and lowest points along each 300 m stream segment.

The BRAT calculates maximum number of beaver dams each stream segment can support, based on the stream power (Qlow and Q2), vegetation, and stream segment slope. Using the calculated maximum number of beaver dams, each stream segment is subsequently classified into the following categories; ‘Pervasive—16 to 40 dams’ (blue), ‘Frequent—5 to 15 dams’ (green), ‘Occasional—1 to 4 dams’ (yellow), ‘Rare—0 to 1 dams’ (orange) and ‘None’ (red). The model also outputs a polyline shapefile of the original drainage network with all the above attribute data added to it and symbology layers to display classified the network in ESRI ArcMap.

Water storage

Beaver ponds in the park were surveyed to determine their water storage. As Riding Mountain National Park is large, and automated beaver pond detection techniques do not yet exist, we manually identified and delineated ponds present on the most recent, high-resolution aerial imagery (2004; 0.625 m resolution; “SI1”). The pond inventory was done within the long-term set of 30 randomly distributed survey blocks, each 24 km2, that the Canadian Wildlife Service established in 1973 to monitor beaver caches. Survey blocks were divided into six equal quadrants (~ 4 km2) and one quadrant per survey block was randomly selected for analysis. The total number and surface area (m2) of beaver ponds in each selected quadrant were calculated using the ‘calculate geometry’ feature in ArcMap 10.5 (see “SI1”). We used the bootstrap technique (10,000 iterations), resampled from the observations, to estimate the beaver pond size distribution for the park. Pond sizes fit a negative exponential distribution—median pond surface area (x) was 1419 m2, and the 95% confidence interval was [1157, 1704 m2] (see “SI2”). Water storage volume (V) was calculated using Karran et al.35

$$V = 0.0955x^{1.1562}$$

for a median-sized beaver pond and its 95% confidence interval. To estimate beaver pond water storage capacity for the entire park, we multiplied the volume of water stored in the median sized pond (421 m3; 95% confidence interval [333, 520 m3]) by the BRAT estimate of beaver dam capacity (see “SI2”).

Changing vegetation composition

To determine whether the vegetation changes over time were real or an artifact of vegetation classification differences in the different years, LANDSAT images from each year were analyzed in ArcMap 10.5 using the time series change detection method Landsat-based detection of Trends in Disturbance and Recovery (‘Landtrendr’) outlined in Kennedy et al.68. The normalized burn ratio (NBR) spectral index was used to determine the disturbance magnitude or cover percentage lost due to disturbance. This approach examines each pixel from year to year to see how it changes over time and highlights it if a substantial change is detected. Additional analysis was undertaken on the highlighted pixels to determine what type of cover change occurred, either a full cover change type or a shift in the spectral index value—indicating cover degradation. Results were compiled into a database to calculate the total percentage of changes. Additionally, results were projected in ESRI ArcMap as shapefiles, to highlight areas of change.

Stream flow changes

Published research predicts significant future stream flow changes in the boreal plains ecozone19,63. To evaluate how climate-induced changes to the streamflow hydrological regime might impact beaver dam capacity, we manipulated the discharge (Qlow and Q2) values in BRAT. Scenarios run included a Qlow increase of 25% and a decrease of 25%. Also, Q2 was increased by 10%, 25% and 100% to simulate the effects of larger 2-year floods. To run the scenarios, we edited the ‘iHyd’ python script in IDLE to include the new discharge rates and ran the BRAT model for each scenario. There were no differences in total dam capacity with the Qlow scenarios as all scenarios resulted in stream powers between 0 and 175 W/m, which BRAT designates as ‘can build dam’. Meaning, decreasing Qlow does not impact the model results, and as a result, BRAT does not recognize any stream reaches being unable to support beaver dams. Thus, reported are only scenarios where Q2 was manipulated.

Data availability

The supplemental information and all non-publicly available data analyzed during this study are available in the following repository:


  1. 1.

    Novakowski, N. S. Population dynamics of a beaver population in northern latitudes. Doctoral dissertation, University of Saskatchewan (1965).

  2. 2.

    Naiman, R. J., Johnston, C. A. & Kelley, J. C. Alteration of North American streams by beaver. BioSci. 38, 753–762 (1988).

    Article  Google Scholar 

  3. 3.

    Jenkins, S. H. & Busher, P. E. Castor canadensis. Mamm. Species 120, 1–8 (1979).

    Article  Google Scholar 

  4. 4.

    Kassi, N. An Indigenous perspective on protecting the Canadian boreal zone. Environ. Rev. 27, 422–423 (2019).

    Article  Google Scholar 

  5. 5.

    Randazzo, M. L. & Robidoux, M. A. The costs of local food procurement in a Northern Canadian First Nation community: An affordable strategy to food security?. J. Hunger Environ. Nutr. 14, 662–682 (2019).

    Article  Google Scholar 

  6. 6.

    Willby, N.J., Law, A., Levanoni, O., Foster, G. & Ecke, F. Rewilding wetlands: beaver as agents of within-habitat heterogeneity and the responses of contrasting biota. Phil. Trans. Royal Soc. B. 373, 2017044. (2018).

    Article  Google Scholar 

  7. 7.

    Rosell, F., Bozser, O., Collen, P. & Parker, H. Ecological impact of beavers Castor fiber and Castor canadensis and their ability to modify ecosystems. Mamm. Rev. 35, 248–276 (2005).

    Article  Google Scholar 

  8. 8.

    Westbrook, C. J., Cooper, D. J. & Butler, D. R. Beaver hydrology and geomorphology. In Treatise on Geomorphology—Ecogeomorphology, (ed. Shroder, J.) 293–306. (Academic Press, San Diego, 2013).

    Google Scholar 

  9. 9.

    Westbrook, C. J., Cooper, D. J. & Baker, B. W. Beaver dams and overbank floods influence groundwater–surface water interactions of a Rocky Mountain riparian area. Water Resour. Res., 42, W06404. (2006).

    ADS  Article  Google Scholar 

  10. 10.

    Karran, D.J., Westbrook, C.J. & Bedard-Haughn, A. Beaver-mediated water table dynamics in a Rocky Mountain fen. Ecohydrology 11, e1923. (2018).

    Article  Google Scholar 

  11. 11.

    Woo, M. K. & Waddington, J. M. Effects of beaver dams on subarctic wetland hydrology. Arctic 43, 223–230 (1990).

    Article  Google Scholar 

  12. 12.

    Nyssen, J., Pontzeele, J. & Billi, P. Effect of beaver dams on the hydrology of small mountain streams: Example from the Chevral in the Ourthe Orientale basin, Ardennes, Belgium. J. Hydrol. 402, 92–102 (2011).

    ADS  Article  Google Scholar 

  13. 13.

    Butler, D. R. & Malanson, G. P. The geomorphic influences of beaver dams and failures of beaver dams. Geomorphology 71, 48–60 (2005).

    ADS  Article  Google Scholar 

  14. 14.

    Polvi, L. E. & Wohl, E. The beaver meadow complex revisited—The role of beavers in post-glacial floodplain development. Earth Surf. Proc. Land. 37, 332–346 (2012).

    ADS  Article  Google Scholar 

  15. 15.

    Puttock, A., Graham, H. A., Cunliffe, A. M., Elliott, M. & Brazier, R. E. Eurasian beaver activity increases water storage, attenuates flow and mitigates diffuse pollution from intensively-managed grasslands. Sci. Total Environ. 576, 430–443 (2017).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  16. 16.

    Wohl, E., Lininger, K. B. & Scott, D. N. River beads as a conceptual framework for building carbon storage and resilience to extreme climate events into river management. Biogeochemistry 141, 365–383 (2018).

    CAS  Article  Google Scholar 

  17. 17.

    Sulla-Menashe, D., Woodcock, C. E. & Friedl, M. A. Canadian boreal forest greening and browning trends: an analysis of biogeographic patterns and the relative roles of disturbance versus climate drivers. Environ. Res. Let. 13, 014007. (2018).

    Article  Google Scholar 

  18. 18.

    Ireson, A. M. et al. The changing water cycle: The boreal plains ecozone of western Canada. Wiley Interdiscip. Rev. Water 2, 505–521 (2015).

    Article  Google Scholar 

  19. 19.

    Hood, G. A. & Bayley, S. E. Beaver (Castor canadensis) mitigate the effects of climate on the area of open water in boreal wetlands in western Canada. Biol. Conserv. 141, 556–567 (2008).

    Article  Google Scholar 

  20. 20.

    Dittbrenner, B. J. et al. Modeling intrinsic potential for beaver (Castor canadensis) habitat to inform restoration and climate change adaptation. PloS ONE 13, e0192538. (2018).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  21. 21.

    Walker, X. J. et al. Increasing wildfires threaten historic carbon sink of boreal forest soils. Nature 572, 520–523 (2019).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  22. 22.

    Thompson, I. D. Habitat needs of furbearers in relation to logging in boreal Ontario. For. Chron. 64, 251–261 (1988).

    Article  Google Scholar 

  23. 23.

    Boulanger, Y. et al. Climate change impacts on forest landscapes along the Canadian southern boreal forest transition zone. Land. Ecol. 32, 1415–1431 (2017).

    Article  Google Scholar 

  24. 24.

    Hood, G. A., Bayley, S. E. & Olson, W. Effects of prescribed fire on habitat of beaver (Castor canadensis) in Elk Island National Park, Canada. For. Ecol. Manage. 239, 200–209 (2007).

    Article  Google Scholar 

  25. 25.

    Fairfax, E. & Small, E. E. Using remote sensing to assess the impact of beaver damming on riparian evapotranspiration in an arid environment. Ecohydrology 11, e1993 (2018).

    Article  Google Scholar 

  26. 26.

    Creed, I. F., Duinker, P. N., Serran, J. N. & Steenberg, J. W. Managing risks to Canada’s boreal zone: Transdisciplinary thinking in pursuit of sustainability. Environ. Rev. 27, 407–418 (2019).

    Article  Google Scholar 

  27. 27.

    Eaton, B., T. Muhly, J.T. Fisher & S-L. Potential Impacts of Beaver on Oil Sands Reclamation Success—An Analysis of Available Literature. Oil Sands Research and Information Network, University of Alberta, School of Energy and the Environment, Edmonton, Alberta. OSRIN Report No. TR-37. (2013)

  28. 28.

    Tape, K. D., Jones, B. M., Arp, C. D., Nitze, I. & Grosse, G. Tundra be dammed: Beaver colonization of the Arctic. Glob. Change Biol. 24, 4478–4488 (2018).

    ADS  Article  Google Scholar 

  29. 29.

    Jarema, S. I., Samson, J., McGill, B. J. & Humphries, M. M. Variation in abundance across a species’ range predicts climate change responses in the range interior will exceed those at the edge: A case study with North American beaver. Glob. Change Biol. 15, 508–522 (2009).

    ADS  Article  Google Scholar 

  30. 30.

    Smeraldo, S. et al. Species distribution models as a tool to predict range expansion after reintroduction: A case study on Eurasian beavers (Castor fiber). J. Nat. Conserv. 37, 12–20 (2017).

    Article  Google Scholar 

  31. 31.

    Scrafford, M. A., Avgar, T., Abercrombie, B., Tigner, J. & Boyce, M. S. Wolverine habitat selection in response to anthropogenic disturbance in the western Canadian boreal forest. For. Ecol. Manage. 395, 27–36 (2017).

    Article  Google Scholar 

  32. 32.

    Sinkins, P. Ecological and hydrological consequences of beaver activity in Riding Mountain National Park, Manitoba. Masters of Science Thesis, University of Manitoba (2008).

  33. 33.

    Trottier, G. C. Aerial Beaver Censuses in Riding Mountain National Park, 1973–1980. Large Mamm. Syst. Stud. Rep. 11, Canadian Wildlife Service (1980).

  34. 34.

    Macfarlane, W. W. et al. Modeling the capacity of riverscapes to support beaver dams. Geomorphology 277, 72–99 (2017).

    ADS  Article  Google Scholar 

  35. 35.

    Karran, D. J., Westbrook, C. J., Wheaton, J. M., Johnston, C. A. & Bedard-Haughn, A. Rapid surface water volume estimations in beaver ponds. Hydrol. Earth Syst. Sci. 352, 1039–1050. (2017).

    ADS  Article  Google Scholar 

  36. 36.

    Wang, G., McClintic, L. F. & Taylor, J. D. Habitat selection by American beaver at multiple spatial scales. Animal Biotel. 7, 10. (2019).

    Article  Google Scholar 

  37. 37.

    Touihri, M., Labbé, J., Imbeau, L. & Darveau, M. North American beaver (Castor canadensis Kuhl) key habitat characteristics: Review of the relative effects of geomorphology, food availability and anthropogenic infrastructure. Ecoscience 25, 9–23 (2018).

    Article  Google Scholar 

  38. 38.

    Whitfield, P. H., Shook, K. R. & Pomeroy, J. W. Spatial patterns of temporal changes in Canadian prairie streamflow using an alternative trend assessment approach. J. Hydrol. 582, 124541. (2020).

    Article  Google Scholar 

  39. 39.

    Johnston, C. A. & Naiman, R. J. Aquatic patch creation in relation to beaver population trends. Ecology 1, 1617–1621 (1990).

    Article  Google Scholar 

  40. 40.

    Whitfield, C. J., Baulch, H. M., Chun, K. P. & Westbrook, C. J. Beaver-mediated methane emission: The effects of population growth in Eurasia and the Americas. Ambio 44, 7–15 (2015).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  41. 41.

    Burchsted, D. & Daniels, M. D. Classification of the alterations of beaver dams to headwater streams in northeastern Connecticut, USA. Geomorphology 205, 36–50 (2014).

    ADS  Article  Google Scholar 

  42. 42.

    Breck, S. W., Wilson, K. R. & Andersen, D. C. Beaver herbivory and its effect on cottonwood trees: Influence of flooding along matched regulated and unregulated rivers. River Res. Appl. 19, 43–58 (2003).

    Article  Google Scholar 

  43. 43.

    Martell, K. A., Foote, A. L. & Cumming, S. G. Riparian disturbance due to beavers (Castor canadensis) in Alberta’s boreal mixedwood forests: Implications for forest management. Ecoscience 13, 164–171 (2006).

    Article  Google Scholar 

  44. 44.

    Hood, G. A. & Bayley, S. E. The effects of high ungulate densities on foraging choices by beaver (Castor canadensis) in the mixed-wood boreal forest. Can. J. Zool. 86, 484–496 (2008).

    Article  Google Scholar 

  45. 45.

    Caners, R. T. & Kenkel, N. C. Forest stand structure and dynamics at Riding Mountain National Park, Manitoba, Canada. Commun. Ecol. 4, 185–204 (2003).

    Article  Google Scholar 

  46. 46.

    Noble, B., Liu, J. & Hackett, P. The contribution of project environmental assessment to assessing and managing cumulative effects: Individually and collectively insignificant?. Environ. Manage. 59, 531–545 (2017).

    ADS  PubMed  Article  PubMed Central  Google Scholar 

  47. 47.

    Thompson, C., Mendoza, C. A. & Devito, K. J. Potential influence of climate change on ecosystems within the Boreal Plains of Alberta. Hydrol. Proc. 31, 2110–2124 (2017).

    Article  Google Scholar 

  48. 48.

    Lapointe St-Pierre, M., Labbé, J., Darveau, M., Imbeau, L. & Mazerolle, M. J. Factors affecting abundance of beaver dams in forested landscapes. Wetlands 37, 941–949 (2017).

    Article  Google Scholar 

  49. 49.

    Hillman, G. R. Flood wave attenuation by a wetland following a beaver dam failure on a second order boreal stream. Wetlands 18, 21–34 (1998).

    Article  Google Scholar 

  50. 50.

    Sallows, T. Beaver abundance—riding mountain dataset. Open Government, Government of Canada. (2018).

  51. 51.

    Petro, V. M., Taylor, J. D., Sanchez, D. M. & Burnett, K. M. Methods to predict beaver dam occurrence in coastal Oregon. Northw. Sci. 92, 278–289 (2018).

    Article  Google Scholar 

  52. 52.

    Graham, H. A. et al. Modelling Eurasian beaver foraging habitat and dam suitability, for predicting the location and number of dams throughout catchments in Great Britain. Eur. J. Wildlife Res. 66, 42 (2020).

    Article  Google Scholar 

  53. 53.

    Richards, L.K. Elk/moose population dynamics in the Riding Mountain National Park region. Masters Thesis, University of Manitoba (1997).

  54. 54.

    Morrison, A., Westbrook, C. J. & Bedard-Haughn, A. Distribution of Canadian Rocky Mountain wetlands impacted by beaver. Wetlands 35, 95–104 (2015).

    Article  Google Scholar 

  55. 55.

    Devito, K. J. et al. Landscape controls on long-term runoff in subhumid heterogeneous Boreal Plains catchments. Hydrol. Proc. 31, 2737–2751 (2017).

    Article  Google Scholar 

  56. 56.

    Westbrook, C. J., Cooper, D. J. & Anderson, C. B. Alteration of hydrogeomorphic processes by invasive beavers in southern South America. Sci. Total Environ. 574, 183–190 (2017).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  57. 57.

    Johnstone, J. F., Hollingsworth, T. N., Chapin, F. S. III. & Mack, M. C. Changes in fire regime break the legacy lock on successional trajectories in Alaskan boreal forest. Glob. Change Biol. 16, 1281–1295 (2010).

    ADS  Article  Google Scholar 

  58. 58.

    Walker, D.J. Landscape complexity and vegetation dynamics in Riding Mountain National Park, Canada. Doctoral dissertation, University of Manitoba (2002).

  59. 59.

    Doucet, R. Regeneration silviculture of aspen. For. Chron. 65, 23–27 (1989).

    Article  Google Scholar 

  60. 60.

    McMaster, R. T. & McMaster, N. D. Composition, structure, and dynamics of vegetation in 15 beaver-impacted wetlands in western Massachusetts. Rhodora 103, 293–320 (2001).

    Google Scholar 

  61. 61.

    Green, K. C. & Westbrook, C. J. Changes in riparian area structure, channel hydraulics, and sediment yield following loss of beaver dams. J. Ecosyst. Manage. 10, 68–78 (2009).

    Google Scholar 

  62. 62.

    Price, D. T. et al. Anticipating the consequences of climate change for Canada’s boreal forest ecosystems. Environ. Rev. 21, 322–365 (2013).

    Article  Google Scholar 

  63. 63.

    Dumanski, S., Pomeroy, J. W. & Westbrook, C. J. Hydrological regime changes in a Canadian Prairie basin. Hydrol. Proc. 29, 3893–3904 (2015).

    Article  Google Scholar 

  64. 64.

    Neumayer, M., Teschemacher, S., Schloemer, S., Zahner, V. & Rieger, W. Hydraulic modelling of beaver dams and evaluation of their impacts on flood events. Water 12, 300 (2020).

    Article  Google Scholar 

  65. 65.

    Westbrook, C. J., Ronnquist, A. & Bedard-Haughn, A. Hydrological functioning of a beaver dam sequence and regional dam persistence during an extreme rainstorm. Hydrol. Proc. 34, 3726–3737 (2020).

    Article  Google Scholar 

  66. 66.

    Andersen, D. C. & Shafroth, P. B. Beaver dams, hydrological thresholds, and controlled floods as a management tool in a desert riverine ecosystem, Bill Williams River, Arizona. Ecohydrology 3, 325–338 (2010).

    Article  Google Scholar 

  67. 67.

    Bailey, R. H. Notes on the Vegetation in Riding Mountain National Park Manitoba. Forest Management Institute (1968).

  68. 68.

    Kennedy, R. E., Yang, Z. & Cohen, W. B. Detecting trends in forest disturbance and recovery using yearly Landsat time series: 1. LandTrendr—Temporal segmentation algorithms. Remote Sens. Environ. 114, 2897–2910 (2010).

    ADS  Article  Google Scholar 

Download references


We thank Cody Misfeldt and Kirby England for field assistance, Thuan Ha for assistance with the vegetation cover change analysis, and the natural resource management staff at Riding Mountain National Park for logistical support. We are grateful to Joe Wheaton who hosted NS for a semester exchange in his research group at Utah State University so she could learn the BRAT model. Funding was provided by Grants from the Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery (RGPIN-2017-05873) and CREATE (463960-2015) programs. This research was carried out under park permit number RMNP-2017-25378.

Author information




C.W. and N.S. designed the research, N.S. performed the research and prepared all figures, and C.W. and N.S. wrote the paper.

Corresponding author

Correspondence to Cherie J. Westbrook.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Stoll, N., Westbrook, C.J. Beaver dam capacity of Canada’s boreal plain in response to environmental change. Sci Rep 10, 16800 (2020).

Download citation


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing