Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2026 Mar 31;32(4):e70833. doi: 10.1111/gcb.70833

Beavers on Ice: Factors Influencing Emergence Phenology in Boreal Canada

Glynnis A Hood 1,✉, D L (Dee) Patriquin 1
PMCID: PMC13036821  PMID: 41914471

ABSTRACT

The influence of climate change on emergence phenology of hibernating mammals in northern latitudes is receiving increased attention, yet for freshwater semi‐aquatic mammals such as beavers, which do not hibernate and therefore maintain an obligatory dependence on freshwater habitats throughout the year, little is known about factors influencing their spring emergence above the ice. Given documented effects of increased warming on patterns of ice formation and ice duration on boreal lakes and ponds, we recorded the date of beaver emergence from their winter lodge onto the ice from 2008 to 2025 (18 years) to examine the possible role of temperature, precipitation, and ice phenology on timing of beaver emergence in the spring. On average, ice‐on to emergence lasted 144 days (min 124.0, max 173.0) with a mean emergence date of April 1st. Increased length of the ice‐free period prior to winter, total precipitation and average temperature from spring ice‐off to the next year's emergence were the best predictors of beaver emergence in spring. Temperature was the best predictor of variation in ice duration, while total precipitation from spring ice‐off to ice‐on and average maximum summer temperature appeared to influence the ice‐on date. Trend analyses revealed increasing annual maximum mean temperatures, with increasing summer temperatures being most apparent. With longer ice‐free periods, beavers can forage longer and store more food for winter access; however, longer periods of open water also increase the period that their activities expose them to increased conflict with humans. Understanding the influence of climate change on the ecology of non‐hibernating, yet ice‐bound mammals provides greater insights into future management of these species.

Keywords: Castor canadensis , climate change, ice phenology, overwintering, semi‐aquatic mammals, trend analysis

Climate‐driven changes in ice phenology influence the timing of spring emergence of a non‐hibernating ice‐bound mammal, the North American beaver.

graphic file with name GCB-32-e70833-g003.jpg

1. Introduction

In more northerly environments, such as Canada's boreal region, temperatures are warming faster than the global average (Woo et al. 2008; Zhang et al. 2021; He and Pomeroy 2023), which can challenge thermoregulatory regimes of mammals. Northern parts of Canada are also experiencing increasingly wetter and warmer weather, while the Prairie provinces to the south face hotter and drier conditions (Singh et al. 2020). These trends suggest that mammalian species occupying higher latitudes and higher elevations might experience more dramatic effects from warming temperatures and changes in precipitation. The directionality of these effects (i.e., positive, negative, no effect) especially during, but not exclusive to, the colder months (Geiser 2020), varies by species relative to the timing and outcomes of seasonal life history strategies (Fietz et al. 2020; Hufnagl et al. 2011; McCain and King 2014).

Investigations into the effects of climate change on wildlife, particularly in higher latitude areas experiencing accelerated warming and increasingly variable precipitation events, reveal mixed results relative to changes in species life‐history traits and population resilience (Wells et al. 2022). Many changes seen in northern mammals (e.g., physiology, reproductive success, survival) are directly associated with the influence of warming temperatures and increased variability in snow conditions on hibernation phenology (Wells et al. 2022; Prather et al. 2023). Some of these changes reflect a phenological mismatch between timing of emergence from winter habitations and timing of spring vegetation growth (Harrington et al. 1999), especially for species that undergo deep torpor throughout the winter. The directionality and form of the phenology response to climate change, however, is not consistent (Geiser 2020).

Positive responses include increased reproduction in edible dormouse ( Glis glis ) with higher ambient spring temperatures and higher food availability (Fietz et al. 2020). A 1°C increase in ambient temperature in spring also advanced the emergence of edible dormice from hibernation by 6 days, but only when food availability was low in the preceding year. Conversely, Turbill and Prior (2016) determined that smaller hibernating rodents experienced a 6.4% decrease in annual survival rate for every 1°C increase in mean annual temperature, with warmer winter temperatures effectively shortening the hibernation period and increasing exposure to predation and other threats. Hufnagl et al. (2011) found a halving of reproductive output for common hamsters ( Cricetus cricetus ) following an exceptionally cold winter. Yet various shrews and many burrowing mammals present no response to changes in hibernation phenology (e.g., McCain and King 2014 and references therein). Mechanisms driving these hibernation strategies appear linked to various factors, including overwintering temperature and metabolic demands, but also spring predation risk and food availability, and preceding factors, such as overwintering food reserves.

Although there is an increased understanding of emergence phenology in hibernating mammals, very little is known about emergence phenology in non‐hibernating mammals that are active but confined over winter (e.g., beavers—Castor spp., common muskrats— Ondatra zibethicus ). As seasonally confined non‐hibernating mammals in northern environments, beaver and muskrat remain active in the lodge and under the ice, while maintaining normothermy throughout the winter. Both North American and Eurasian beavers ( Castor canadensis and C. fiber ) in northern latitudes are confined to their lodges/dens and under‐ice habitats in winter waterbodies for up to 8 months or more (Aleksiuk 1970). Unlike muskrats, which forage for submerged aquatic vegetation under the ice throughout the winter (Lorentz and Hood 2025), in northern climates, beavers build and store a substantial food cache of woody stems underwater at one or more entrances to their winter lodge or bank den, which is then accessed throughout winter (Busher 1996). Beavers also supplement their winter diet with roots of submerged aquatic vegetation accessed from under the ice (Jenkins and Busher 1979; Milligan and Humphries 2010).

Despite subzero temperatures throughout the winter, temperatures within the lodge remain above or near 0°C due to the insulative nature of the lodge and body heat produced by the beaver family (Miller 1967 in Baker and Hill 2003; Stephenson 1969). Aleksiuk and Cowan (1969) identified a slight metabolic depression in Arctic beavers following light‐exposure experiments comparing beavers from California and those originating from the Arctic, and a study in northern Minnesota and upper Michigan recorded a slightly reduced core body temperature (Tb) for adults and even less for juveniles versus kits over winter (Smith et al. 1991); however, there is no evidence that beavers experience torpor or dramatic changes in body temperature during winter, especially when not foraging under the ice (Dyck and MacArthur 1992; Reynolds 1992).

Many hibernating mammals often mate after emergence (e.g., hoary marmot Marmota caligata , Columbian ground squirrel Urocitellus columbianus ), while others mate in autumn but experience delayed implantation while in winter confinement (e.g., American black bear Ursus americanus ). Gonadal atrophy associated with hibernating mammals in deep torpor and mammals with daily torpor (daily heterotherms) prevents breeding while in hibernation (Bieber et al. 2018). Conversely, beavers in northern latitudes and montane regions are normothermic and mate in the lodge in mid to late winter (January to March). After 98–111 days, the pair produces one annual litter comprised of two to three kits, on average (Baker and Hill 2003). Beavers are socially monogamous, with some exceptions (Mayer, Künzel, et al. 2017), and live in a family group comprised of adults, juveniles, and young of year. Young beavers remain with the family for approximately 2 years before dispersal, although dispersal can be delayed in areas with high population densities (Sun et al. 2000; Mayer, Zedrosser, and Rosell 2017). The highly social nature of beavers aids thermoregulation inside the lodge. As with hibernators and daily heterotherms, beavers also experience a distinct timing of post‐winter emergence in spring, when they first appear in a newly formed, small area of open water in front of their lodges, eating woody stems from the underwater and ice‐bound parts of the food cache (Bromley and Hood 2013). Open water adjacent to the main entrance to the active lodge occurs well before the rest of the pond is ice‐free, aided in part by thinner ice in this area (Hood, unpublished data). Beavers are observed in this opening soon after it opens (“spring emergence”) and the same opening does not appear at inactive lodges (Bromley and Hood 2013). Beavers maintain a level of confinement to the lodge, and the water and ice next to the lodge for several days or even weeks, until ice melts at the pond edge and allows access to other food resources (e.g., cattail roots).

The influence of climate change on emergence phenology of various hibernating mammals in northern latitudes is receiving increased attention, yet for freshwater semi‐aquatic mammals such as beavers, which do not hibernate and therefore maintain an obligatory dependence on freshwater habitats throughout the year, little is known about factors influencing the timing of their spring emergence above the ice, or effects of climate change. The dependence of beavers on stored winter food, continued under‐ice activity over winter, and within‐lodge reproduction during winter sets them apart from traditional studies of hibernating mammals (see reviews by Wells et al. 2022; Findlay‐Robinson et al. 2023).

In this study we examine beaver emergence phenology over an 18‐year period (January 2008 to December 2025) at a large beaver pond in Canada's southern boreal forest. Given that beaver pond occupancy is often much shorter in this study area, on average 3.62 years from previous studies (Hood 2020b), the continuous pond occupancy by beavers paired with matched ice‐phenology data at this site provides a unique opportunity to study this phenomenon over time. Our goal was to investigate which climatic variables influence spring emergence of North American beavers in the small opening near the lodge, and above the ice from their winter lodges. Our objectives were to (1) assess climatic variables over the same period for which we had beaver emergence and ice phenology data, (2) track the date of first emergence of beavers above the ice in the spring over successive years, and (3) record ice‐on and ice‐off dates over this same period. We hypothesized that warmer average winter temperatures (i.e., during the overwinter ice‐on to ice‐off period) would result in earlier emergence dates of beaver, while colder average winter minimum temperatures coupled with low precipitation would delay emergence dates due to thicker ice formation. We also hypothesized that increased overwinter precipitation would result in earlier emergence dates due to increased insulation of ice, which would lead to thinner overall ice. Lastly, we hypothesized that the increased period from spring ice‐off to ice‐on would result in later beaver emergence dates because of longer summer foraging and fall caching periods. We have observed beavers chewing and breaking ice to access the upper cache stems held in ice at the time of emergence, which suggests diminished winter food resources may play a role in emergence.

Although research addresses the effects of climate change on hibernating mammals, there is a paucity of research on how shifts in temperature and precipitation relative to climate change might affect emergence phenology and foraging periods of a non‐hibernating, yet ice‐bound mammal. No semi‐aquatic mammal, including beavers, hibernates, despite having ranges that include long, cold winters that are increasingly impacted by climate change (Hood 2020a). Their dependence on ice conditions and ability to forage under the ice is fundamental to overwinter survival. Increasing our understanding of potential thresholds and influences in their emergence phenology is essential as beavers continue to expand their northern range in response to a warming climate, which consequently alters these novel landscapes in unpredictable ways (Tape et al. 2018).

2. Methods

2.1. Study Area

The study area is within Miquelon Lake Provincial Park (MLPP, 13 km2, 53.2363°N, 112.9068°W; Figure 1) in east‐central Alberta, Canada, which is part of a morainal landscape within the southern dry mixed‐wood boreal forest. The park has a humid continental climate (Köppen climate classification Dfb) with warm summers and cold winters that regularly sustain temperatures below −20°C. On average, winter snowfall accumulations are between 100 and 120 cm. There are numerous pothole wetlands in the park surrounded by riparian forests dominated by trembling aspen ( Populus tremuloides ) and willow (Salix spp.). Beavers are common in the park and most of the park's wetlands have been occupied by beavers at least once (Hood 2020b). There are no rivers in the park, and streams are intermittent; therefore, beaver dams are rare.

FIGURE 1.

FIGURE 1

Open in a new tab

Beaver lodges and ponds in Miquelon Lake Provincial Park, Alberta, Canada including Grebe Pond (purple).

We have conducted surveys of winter beaver lodge occupancy and pond condition on all ponds in the park since 2008 (Bromley and Hood 2013; Hood 2020b). At 9.2 ha, Grebe Pond (Figure 1) is one of the largest ponds in MLPP and is the only pond in the park consistently occupied by beavers, assumed to be the same family group, since 2008 (i.e., parents and generational offspring). When surveys started in 2008, the pond had two beaver lodges. Another lodge was built in 2009 and the fourth one in 2024. All four lodges have been occupied at least once since 2008, with beavers in Grebe Pond switching within‐pond lodge residency an average of every 1.38 years (SD = 0.52 years, minimum = 1 year, maximum = 2 years; Hood 2020b). On average, other ponds that had some beaver occupancy since 2008 (n = 56) in the park were continuously occupied for 2.83 years (SD = 2.21 years; Hood 2020b). As with many of the waterbodies in MLPP, Grebe Pond is shallow, with a maximum depth just over 2.5 m. Its perimeter is 4012 m and, lacking connectivity with other waterbodies, all water comes from precipitation and groundwater. Given the uninterrupted 18‐year occupancy of Grebe Pond by beavers, our study focuses exclusively on the timing of spring emergence by beavers in this pond.

2.2. Data Collection

2.2.1. Beaver Surveys

With the help of volunteers, we have surveyed every beaver lodge in MLPP (n = 139) in early winter since January 2008 (Bromley and Hood 2013) to document the occupancy status of each lodge (active/inactive), lodge type (bank/island/den), lodge condition (1 = maintained and occupied, 2 = maintained but unoccupied, 3 = unoccupied but roof intact, 4 = unoccupied and roof collapsed, 5 = only sticks remaining, 6 = sunk/not found). Any new lodges are also mapped and added to the dataset after each annual survey. Since 2010, we have also recorded pond condition (1 = full of water to the edges, 2 = muddy shoreline exposed, 3 = water in beaver canals only, 4 = dry pond/meadow). The authors have personally conducted the occupancy surveys on Grebe Pond for the entire 18‐year period. Grebe Pond is the only pond in our study area that was continuously occupied through this period; average occupancy for all park lodges is 3.62 years (Hood 2020b). Continuous occupancy in a single pond offers a unique opportunity to investigate climate influences on emergence, while controlling other potential influences (e.g., pond size and depth, available forage).

Beaver lodges are considered “active” when a winter food cache is present, the lodge is maintained, and when temperatures are cold enough, there is an obvious frost vent at the top of the lodge. Additional evidence of recent beaver activity is also noted (e.g., freshly cut stems nearby or on the lodge). To quantify drying trends throughout the park, we used pond condition values from all beaver ponds in MLPP, including ones that had an abandoned lodge but had not been occupied since occupancy surveys began in 2008 (n = 86) to derive a mean score for each year since these values were recorded (2010–2024). There was not enough variability just for Grebe Pond (either a pond class 1 or 2, but no lower) to assess the overall impact of precipitation and temperature on water levels in the overall system.

2.2.2. Climate

We obtained climate data from the Alberta Climate Information Service's online portal (https://www.alberta.ca/acis‐find‐historic‐climate‐data). We downloaded daily weather data from 2007 to present, including: daily mean temperature (°C), average daily minimum daily temperature, daily maximum temperature, daily precipitation (mm), and daily average relative humidity (%).

Next, we calculated the mean temperature, relative humidity, and total precipitation values for the days between the various combinations of emergence dates, ice‐on dates, and ice‐off dates (Table 1; e.g., average temperature from ice‐on date to the subsequent emergence date). Emergence day, ice‐off day, and ice‐on day dates were standardized to the number of days since 1 January for analysis of climatic factors influencing these dates. Time intervals included: average temperature 1 month prior to emergence day, average temperature (and minimum and maximum temperatures), relative humidity, and total precipitation from emergence day the year prior to the next emergence day. The same calculations were used for the intervals between the ice‐off day of year (“ice‐off”) to the next emergence day, ice‐on to the following ice‐off day of year, ice‐off to the next ice‐on day, emergence day to ice‐on, and previous emergence day (t‐1) to the next ice‐off day.

TABLE 1.

Predictor variables for ice‐on date ice duration and linear modeling and AICc model selection for (1) emergence day of beavers, (2) ice‐on day, and (3) ice duration on Grebe Pond in Miquelon Lake Provincial Park, Alberta Canada.

Response variable Collinear predictors Other predictors
Emergence day

  • Average temperature from ice‐off to emergence

  • Average minimum temperature from ice‐off to emergence

  • Average maximum temperature from ice‐off to emergence

  • Average temperature 1 month prior to emergence

  • Total precipitation from ice‐off to emergence

  • Total precipitation from ice‐on to emergence

  • Number of days between ice‐off and ice‐on the season prior to emergence
Ice‐on day

  • Average temperature 15 May–30 September (Summer average temperature)

  • Total precipitation from ice‐off to ice‐on

  • Average maximum temperature 1 month prior to ice‐off

  • Average relative humidity 15 May—30 September (Summer average RH)
Ice duration (ice‐on to ice‐off)

  • Average temperature from ice‐on to ice‐off

  • Average minimum temperature from ice‐on to ice‐off

  • Average maximum temperature from ice‐on to ice‐off

  • Average temperature 1 month prior to ice‐off

  • Total precipitation from ice‐on to ice‐off

  • Average relative humidity from ice‐on to ice‐off

  • Average minimum temperature from ice‐on to ice‐off × Total precipitation from ice on to ice‐off

  • Average temperature from ice‐on to ice‐off × Average relative humidity from ice‐on to ice‐off
Open in a new tab

For trend analysis of temperature and precipitation over the 18‐year period, daily values were used. Annual calculations for temperature modeling started from 1 April to 31 March to reflect a year of data from the average emergence day (1 April). Non‐ice‐bound months were from 1 April to 30 September, and ice‐bound months were from 1 October to 31 March. To examine seasonal temperature trends over time, data were in the following categories: “Spring” = April to June, “Summer” = July to September, Fall = October to December, and “Winter” = January to March.

2.2.3. Emergence Phenology

Every year, we checked the open‐water area on first appearance and recorded the first day that beavers emerged from their lodge on Grebe Pond into the small area of open water that forms in front of the lodge in spring (“emergence day”). This open‐water area appears well before the rest of the pond, including pond edges, begins to melt, thereby restricting beaver activity to that small space. They do not move to other parts of the pond until open leads form in the ice, sometimes 2–4 weeks later. Adult beavers are usually observed first in the open water, followed by other family members after several days. The adults are obvious in the water and on the adjacent ice, typically at dusk, and we checked for activity in afternoon through dusk each day after the open water appeared until the first activity was observed. As noted above, we standardized the date to the number of days since January 1 of that year. We did not begin formal beaver lodge occupancy surveys until January 2008; therefore, the emergence date for 2007 is unknown, which prevented the calculation of some variables for 2008 (e.g., time from ice on in 2007 to emergence date in 2008).

2.2.4. Ice Conditions

Every year, we recorded ice‐on dates, when the pond was completely covered in ice (including immediately in front of the lodge) and ice‐off dates, when the pond was completely free of ice. For some years, ice conditions were missed due to brief absence from the study area. The ice‐off date for 2008 was estimated to be April 6th from personal photographs taken at the study area and the ice‐on date for 2008 was obtained from local fishing reports. For the ice‐off date for 2018 and the ice‐on date for 2022, we used aerial imagery obtained from the Copernicus Browser in the Sentinel Hub (https://www.sentinel‐hub.com/). Given the quality of the image sources, we are confident that we accurately captured the true dates to ±1 day. Because we did not have ice condition data for 2007 and suitable Copernicus images were not available for this timeframe, we were unable to calculate some variables for 2008 (e.g., time from ice‐on in 2007 to emergence date in 2008). We calculated the duration of the ice‐free period for each year (foraging period for beavers), ice duration over the winter (ice‐on to ice‐free date), and time under ice for beavers (ice‐on to emergence date).

2.3. Data Analyses

2.3.1. General

We conducted all analyses in R (4.5.0, R Core Team 2025). All data were tested for normality using the Shapiro–Wilk test and Q–Q plots (visual assessment), and where appropriate, the Levene's test to test for homoscedasticity. All data were considered significant at α = 0.05. When normality could not be met, non‐parametric approaches were used. For linear regression modeling, we used the following R packages: MASS for stepwise model selection using Akaike Information Criterion (AIC), car to quantify multicollinearity using the Variance Inflation Factor (VIF), AICcmodavg for AICc (second‐order AIC model selection), tidyverse to help transform and present data, and MuMIn (Multi‐Model Inference) for automated model selection and model averaging. For the 18‐year period, we also calculated descriptive statistics (mean, SD, minimum, maximum) for emergence day, ice‐on day, ice‐off day, length of ice‐free period, ice duration, days between ice‐on date and emergence date, days between emergence and ice‐off date, and pond condition (since 2010).

2.3.2. Climate Trends—Temperature

In R, we used the packages forecast, ggplot2, lubridate, tidyverse, and tseries to develop autoregressive integrated moving average (ARIMA) models (Lai and Dzombak 2020) as a time series forecasting technique for daily temperature data from 1 April 2007 to 1 April 2025. These data represent a year between the average emergence date for beavers in the study pond (“beaver year”), the ice‐free period (1 April to 30 September), and the ice‐on period (1 October to 31 March). As a statistical technique, the ARIMA model integrates the most recent observations with long‐term historical trends to help identify “breaks” in time series patterns and measures goodness‐of‐fit relative to the sample mean (Lai and Dzombak 2020). Prior to analyses, we tested the daily temperature data for linearity, independence, homoscedasticity (Levene's test) and normality (Shapiro–Wilks test and Q–Q plots for visual assessment).

For ARIMA modeling, we first graphed temperature trends over time and then graphed decomposition of additive time series followed by an Augmented Dickey‐Fuller Test to determine if the time series was stationary or non‐stationary (Dickey and Fuller 1979). Specification for the annual ARIMA models (1 April to 31 March) were (3,0,0), indicating 3 AutoRegressive (AR) terms. These terms are past values of the time series that were used to predict the future values; 0 differencing, which assumes a stationary time series (no need to use differencing to stabilize the mean of the time series), and 0 moving average (MA), because past forecast errors were not used to make predictions. Additionally, we set the model to have 0 seasonal AR terms, 1 for seasonal differencing (removes seasonality because the data were with a period of 365 days), and 0 seasonal MA terms. We set the seasonal cycle to 365 periods (days in a year) to allow for modeling over a year.

For temperature trends in the period following average emergence to the start of winter (1 April—1 October), only data for this period were included in the ARIMA model. Specifications of this model were (2,0,3), indicating two AR terms, meaning that the next predicted value in the time series depended on its values at the previous two time steps; zero differencing, and three moving average (MA) terms to account for residual errors from the previous three time steps. For the seasonal part of the model, we set the model to have zero AR terms, one order of differencing, and no MA terms. The length of seasonal cycle was 182 days to represent the half‐year cycle in the data. The same approach was used for the period when beavers were generally confined to the pond for the winter (1 October—31 March). Finally, we ran ARIMA to graph and forecast temperature outcomes based on the historical time series.

2.3.3. Climate Trends—Precipitation

To assess annual trends in precipitation, we first defined a custom “water year” (365 days) starting on 1 April, the average emergence date for beavers. We then calculated total precipitation by year and analyzed the trends using a linear regression model with total precipitation as the response variable and water year as the predictor variable. Next, we analyzed the data for seasonal trends, using the “Spring” (April—June), “Summer” (July to September), “Autumn” (October to December), and “Winter” (January to March) time periods. Finally, we visually compared seasons across years. To detect monotonic trends in each season, we performed a non‐parametric Mann‐Kendall Trend Test (Mann 1945; Kendall 1975) using the Kendall package in R. This test assessed whether values increased or decreased by season. Finally, we used a one‐way ANOVA to test for differences in seasonal precipitation. Any significant results were explored using a Tukey's HSD test. Results were significant at α = 0.05.

2.3.4. Emergence Date

To quantify the effects of temperature, precipitation, and ice phenology on the timing of emergence day over the 18‐year period, we used the MuMIn package in R to create linear models of all possible combinations of predictors using the power set formula to identify all possible subsets (n = 28 models). The four temperature variables from ice‐off to ice‐on (average temperature from ice‐off to emergence date, average minimum temperature from ice‐off to emergence date, average maximum temperature from ice‐off to emergence date, and average temperature 1 month prior to emergence) were set as collinear variables to ensure they never appeared together in the same model. We then defined the other predictors (Table 1). Once the models were calculated, we also assessed multicollinearity by calculating a variance inflation factor (VIF) for each variable within each model. Any variables with VIF > 3 were excluded from the model in which multicollinearity was evident (Dormann et al. 2013).

To identify the best‐fit models to predict emergence day over the 18‐year period, we used an information‐theoretic approach with AICc to evaluate competing linear models to explain interannual variation in emergence dates. Any models with a ΔAIC < 2 represented the most parsimonious models (Burnham and Anderson 2002). Beta coefficients (β), standard errors (SE), and confidence intervals (CI) for the predictors were calculated using a model selection approach and represent estimates conditional on the selected model. To ensure all assumptions for linear modeling were met, we created normal Q–Q plots and a histogram of the residuals to test whether the residuals were normally distributed. Then we tested homoscedasticity with scale‐location plots and used Cook's Distance to identify possible influential data points.

Ice phenology: As with emergence data, we created models of all possible combinations of predictors for ice‐on day (n = 15 models; Table 1) and ice duration (n = 92 models; Table 1). An appropriate date range for temperature data for the period between ice‐off and ice‐on (ice‐free period) was difficult to determine, since it encompassed three seasons (spring, summer, and autumn). Complicating things further, ice‐on day was recorded for the date that Grebe Pond was completely ice covered; however, the pond could be almost completely covered with ice for several days or even weeks before all water was frozen on the entire pond and the criterion for ice‐on day was met. Of note, water directly in front of an occupied beaver lodge is always the last to freeze and first to thaw (Bromley and Hood 2013). Air temperatures during autumn play a role in ice formation, along with wind and precipitation (Fujisaki‐Manome et al. 2020). Given this pattern of freeze‐up, we used temperatures for ice‐on day instead of the whole period from ice‐off to ice‐on to model timing of ice‐on, and temperature metrics from 15 May 15 to 30 September to separate the influence of spring through autumn temperatures on ice‐on.

For models assessing the ice‐bound period, permanent ice cover (“ice duration”) was fully within the winter months and temperature variables were not adjusted. For the ice duration models, we created two interaction terms: (1) Average minimum temperature from ice‐on to ice‐off × Total precipitation from ice‐on to ice‐off (to accommodate snow accumulation) for snow insulation effects, and (2) Average temperature from ice‐on to ice‐off × Average relative humidity (RH) from ice‐on to ice‐off, because of the role RH plays in ice formation (Duguay et al. 2003). Once the candidate models were generated, we followed the same procedures and thresholds described previously for the emergence data.

3. Results

3.1. Temperature and Precipitation

There was considerable variability in temperature and precipitation in the study area since 1 April 2007 (Table 2). Average daily, minimum, and maximum temperatures and total precipitation fluctuated, depending on the year (Table S1).

TABLE 2.

Average, maximum, and minimum temperatures and total annual precipitation from 1 April 2007 to 1 April 2025 in Miquelon Lake Provincial Park, Alberta. ‘Value’ indicates the maximum or minimum value for each of the column headings over the 18‐year period.

Value Average daily temperature Minimum daily temperature Maximum daily temperature Total precipitation
Maximum 26.1°C (30 June 2021) 19.0°C (23 July 2024) 35.4°C (30 June 2021) 513 mm (2016/2017)
Minimum −38.2°C (13 January 2024) −44.5°C (13 January 2024) −32.6°C (12 January 2024) 228 mm (2009/2010)
Open in a new tab

For average daily temperature, minimum daily temperature, and maximum daily temperature from 1 April 2007 to 1 April 2025, the ARIMA models were generally well‐specified with significant autoregressive (AR) terms (p < 0.05), indicating that the past temperatures (e.g., the previous day's temperature) provided useful information for predicting the next day's temperature. For all three models (annual ARIMA model, beaver active period, and confined period), significance of the augmented Dickey–Fuller test for all the models (p = 0.01) provided strong evidence against H 0 (unit root), indicating that the data were likely stable over time.

Trend analyses for annual mean daily temperature and annual minimum daily temperature were not significant, while the trend for annual maximum daily temperature revealed a trend towards increasing maximum temperatures, with increasing temperatures predicted in future (F 1,6574 = 3.27, p = 0.07; Figure 2). The model for annual maximum temperature revealed a reasonably good fit with relatively low root mean squared error (RMSE = 5.91), mean absolute error (MAE = 4.45), and mean absolute scaled error (MASE = 0.6053).

FIGURE 2.

FIGURE 2

Open in a new tab

Temperature trends over time for daily maximum temperature (2007/2008–2024/2025) graphed as the decomposition of additive time series for the following components: Random (residuals), seasonal (periodic or regular fluctuations that repeat at fixed intervals), trend (overall direction in the data over time), and observed (actual data observed).

When the data were separated into distinct half‐year periods to account for seasonality, trends became more apparent. For the period when beavers are generally active (1 April—30 September), there was a strong trend for warmer average daily temperatures (F 1,3292 = 13.51, p = 0.0002), minimum daily temperatures (F 1,3292 = 16.21, p < 0.001), and maximum daily temperatures over time (F 1,3292 = 9.04, p < 0.003; Figure 3). For the period when beavers are generally confined to their lodge and under‐ice habitat over winter (1 October—31 March), no strong trends emerged from the data.

FIGURE 3.

FIGURE 3

Open in a new tab

Temperature trends over time for average daily temperature (A), minimum daily temperature (B), and maximum daily temperature (C) (1 April 2007/2008–30 September 2024/2025) graphed as the decomposition of additive time series for the following components: Random (residuals). Seasonal (periodic or regular fluctuations that repeat at fixed intervals), trend (overall direction in the data over time) and observed (actual data observed).

Annual and seasonal precipitation (1 April—31 March) were variable over the 18‐year study period (Figure 4). Despite finding no significant trend within seasons (Mann–Kendall tau p > 0.05), there was a distinct seasonal difference in total precipitation among seasons (F 3,68 = 76.87, p < 0.001; Figure 5). Differences among seasonal means indicate that spring and summer had more total precipitation than autumn and winter (Table 3).

FIGURE 4.

FIGURE 4

Open in a new tab

Total annual precipitation (A) and total seasonal precipitation (B) trends for 1 April–31 March (Water Year) from 2007 to 2025. For total annual precipitation (A) the red line represents the trend line, gray shading indicates the confidence interval around the regression line.

FIGURE 5.

FIGURE 5

Open in a new tab

Distribution of precipitation by season in Miquelon Lake Provincial Park, Alberta, Canada (2007–2025).

TABLE 3.

Mean differences, 95% confidence intervals, and significance of total seasonal precipitation by season from 2007 to 2025 (Tukey HSD, α = 0.05) at Miquelon Lake Provincial Park, Alberta, Canada.

Comparison Mean difference (mm) Confidence interval (mm) p
Spring‐Autumn 108.6 79.0, 138.2 < 0.001
Summer‐Autumn 125.5 95.9, 155.1 < 0.001
Winter‐Autumn −6.1 −35.6, 23.6 0.95
Summer–Spring 16.9 16.9, −12.7 0.44
Winter–Spring −144.6 −144.2, −85.0 < 0.001
Winter‐Summer −131.5 −161.1, −101.9 < 0.001
Open in a new tab

3.2. Emergence Dates

Average emergence date since 1 January for beavers from 2007 to 2025 was 90 days (1 April, SD = 11.6 days), with a minimum of 66 days (16 March 2016) and maximum of 110 days (19 April 2020). On average, beavers were confined under the ice for 144 days (SD = 14 days), with a minimum of 123 days and a maximum of 172 days.

For all linear models, VIF was < 3; therefore, multicollinearity was not an issue among predictors. The top model for emergence date included total precipitation from ice‐on to emergence day and average temperature from ice‐off the previous year to emergence (Table 4). Total precipitation from ice‐off to emergence in the top‐ranked model had a weak but positive effect on emergence day (β = 0.25, SE = 0.16, 95% CI = −0.095, 0.61), although a 95% CI overlapping zero indicated a possible unidentified interaction between the predictor variables (Table 5). In the second ranked model, for every 1°C decrease in average temperature from ice‐off in the previous year to emergence, emergence day increased by 6.0 days (95% CI = −10.6, −1.4). Specifically, emergence day increased with lower temperatures in the previous year (Table 5).

TABLE 4.

Results of top individual model runs for linear models using a Gaussian distribution of spring emergence day for beaver ( Castor canadensis ) over 18 years (from 2008 to 2025) in Miquelon Lake Provincial Park Alberta, Canada, where df is degrees of freedom, AICc is Akaike's information criterion corrected for small samples size, and ω is model weight. Models with ΔAICc < 2 are considered to have strong support (Burnham and Anderson 2002).

Candidate model df log‐likelihood AICc ΔAICc ω
Total precipitation from ice‐on to emergence + Average temperature from ice‐off from the previous year to emergence 4 −60.03 139.4 0.000 0.26
Total precipitation from ice‐off to emergence + Average temperature from ice‐off in the previous year to emergence 4 −60.24 131.8 0.423 0.21
Open in a new tab

TABLE 5.

Coefficients, standard errors, and 95% confidence intervals for of top individual model runs for linear models using a Gaussian distribution of spring emergence day for beaver ( Castor canadensis ) over 18 years (from 2008 to 2025) in Miquelon Lake Provincial Park Alberta, Canada.

Model Term Estimate SE Lower 95 Upper 95
M1 Intercept 88.96 14.22 58.46 119.46
M1 Total precipitation from ice‐on to emergence 0.26 0.16 −0.10 0.61
M1 Average temperature from ice‐off the previous year to emergence −4.34 2.50 −9.691 1.01
M2 Intercept 91.75 13.59 62.60 120.90
M2 Total precipitation from ice‐off to emergence 0.043 0.15 −0.02 0.11
M2 Average temperature from ice‐off the previous year to emergence −5.97 1.08 −10.56 −1.39
Open in a new tab

3.3. Ice Phenology

The average ice‐on date was 313 days from 1 January (12 November; SD = 10 days). The earliest ice‐on date was 298 days from 1 January (25 October 2020), and the latest ice‐on date was Day 328 (24 November 2025). On average, the pond was completely ice‐free on Day 112 (22 April; SD = 9 days). The earliest that the pond was completely ice‐free was on Day 95 (5 April 2016) and the latest was Day 124 (4 May 2013 and 2014). For ice duration, the average consecutive ice cover lasted 164 days (SD = 11 days), with a maximum duration of 181 days (2013/2014) and minimum duration of 144 days (2023/2024). The average number of days from beaver emergence to ice‐off was 20 (SD = 9 days), with a maximum of 34 days (27 March 2017) and minimum of 7 days (26 April 2020).

3.3.1. Ice Duration

As with emergence models, VIF was < 2 for all linear models involving ice phenology; therefore, multicollinearity was not an issue among predictors. Temperature variables were the best predictors of the length of ice duration in Grebe Pond (Table 6), where average temperature 1 month prior to ice‐off and average maximum temperature from ice‐on to ice‐off represented the top model, although with all model weights 1.0, the top model was not notably stronger than the other candidate models. In all the top models, the confidence interval for the predictor variables overlapped zero, which might indicate an unidentified interaction or multicollinearity that was not identified by VIF among predictors for these models (Figueiras et al. 1998). None of the interaction terms appeared in any of the top models.

TABLE 6.

Results of top individual model runs for linear models using a Gaussian distribution of ice duration and ice‐free period from 2008 to 2025 in Miquelon Lake Provincial Park Alberta, Canada, where df is degrees of freedom, AICc is Akaike's information criterion corrected for small samples size, and ω is model weight. Models with ΔAICc < 2 are considered to have strong support.

Period Candidate model df Log‐likelihood AICc ΔAICc ω
Ice duration Average temperature 1 month prior to ice‐off + Average maximum temperature from ice‐on to ice‐off 4 −63.11 137.5 0.000 0.09
Average temperature 1 month prior to ice‐off + Average minimum temperature from ice‐on to ice‐off 4 −63.2 137.8 0.213 0.08
Average temperature 1 month prior to ice‐off + Average temperature from ice‐on to ice‐off 4 −63.3 138.0 0.473 0.07
Average minimum temperature from ice‐on to ice‐off × Total precipitation from ice‐on to ice‐off + Average maximum temperature from ice‐on to ice‐off 4 −63.7 138.7 1.120 0.05
Ice‐on day Average maximum temperature 1 month prior to ice‐on 3 −65.1 138.0 0.000 0.32
Average summer temperature 3 −65.5 139.3 0.489 0.27
Average maximum temperature 1 month prior to ice on + Average summer temperature 4 −64.4 139.5 1.84 0.13
Open in a new tab

3.3.2. Ice‐On Date

The average maximum temperature 1 month prior to the ice‐on date represented the top model predicting ice‐on date, while summer average temperature was also an important variable (Table 6). Both higher average temperatures for the 1 month prior to the ice‐on date (β = 1.4, SE = 1.7) and average summer temperature (β = 1.5, SE = 3.0) resulted in later ice‐on dates. As with the models for ice duration, confidence intervals for all models overlapped zero, suggesting that other factors also play a role in the timing of ice formation.

4. Discussion

Behavioral, morphological, and biogeographic traits in many species can influence their ability to counter the impacts of climate change; however, the magnitude and seasonal timing of these changes can test the adaptive ability of species or populations to sustain themselves. For mammals living in temperate and more northerly regions, the effects of warming temperatures are of concern given the seasonal nature of their life history traits. With predicted increases of 4°C–5°C in mean annual temperatures across the boreal zone in Canada by 2100 (Price et al. 2013), understanding current and potential future changes in seasonally‐driven ecosystems and the species on which they depend is paramount.

Given current warming trends, we hypothesized that there would be an increase in average winter temperatures over the study period, thus resulting in earlier emergence dates for beavers over time; however, no strong trends emerged for winter temperatures. Yet, during the period when beavers are generally free to move and ice was either thawing rapidly or gone (1 April–30 September), daily maximum temperatures increased year over year. This trend held for all temperature metrics (average, minimum, and maximum daily). Higher temperatures during the foraging period would slow ice formation and, therefore, allow for more time for beavers to build larger food caches and access fresh forage to increase winter fat storage. Hibernating mammals, the closest comparator to winter confinement of beavers under the ice, experience similar benefits from extended foraging opportunities, particularly if emergence is timed with earlier plant growth (Macphie and Phillimore 2024).

The timing of winter confinement appeared most influenced by climatic variation, particularly for ice‐on and ice‐off timing. Beavers at MLPP spend between 4.1 and 5.8 months of the year confined to their lodge and under‐ice habitat. Immergence occurred immediately upon freeze‐up of all leads from the lodge to shore, which usually corresponded to freezing of the remaining open water in front of the lodge. This area is always the last to freeze and first to melt (Bromley and Hood 2013). In Massachusetts, USA, which is much further south than our study area (42o35′N, 72o40′W), Lancia et al. (1982) determined that beaver activities above the ice were absent below −10°C. Despite fluctuating temperatures during freeze‐up, we did not identify this activity trend but, unsurprisingly, found that warmer summer and fall temperatures were associated with later ice‐on dates at MLPP. Once beavers emerge for the first time in spring, they are still restricted to the open‐water area in front of the lodge for several days, if not weeks before they were able to forage further afield when shoreline ice had melted. Indeed, we observed that re‐freezing events in early spring with incoming cold fronts, even after emergence, caused beavers to remain confined to the small area of water in front of their lodges for several days longer than other years, thus delaying any access to supplemental foraging opportunities.

For example, in 2025 beavers first emerged from their lodge on 24 March; however, a cold spell ensued, and the ice‐off date for the entire pond was not until 18 days later (14 April). During this time, beavers consumed materials from their food cache and, as the ice became more friable and gradually thinned around the food cache, beavers were able to access woody stems that had been otherwise trapped in the ice over winter. Relying on woody stems in early spring immediately following emergence is not unusual; however, unlike with Aleksiuk (1970) who observed Arctic beavers cutting saplings along the shore and moving them under the ice, beavers in Grebe Pond did not go to shore until open leads in the ice allowed for safe passage. This difference could be because of the lotic systems used by beavers in Aleksuik's study area versus the completely lentic nature of ponds in MLPP; increased spring water flow along river systems and the addition of overland flow from freshet may facilitate ice melt adjacent to shorelines, allowing earlier access to terrestrial food resources. Alternatively, colder average temperatures in the Arctic required beavers to forage beyond the lodge in late winter and early spring because of depleted food cache resources (Aleksiuk 1970).

Emergence dates were primarily influenced by increased temperatures over the winter period and had a weak but positive relationship with precipitation, which together we surmised would create synergistic reactions influencing ice thickness (Murfitt et al. 2018; Fujisaki‐Manome et al. 2020). Indeed, positive interactions among weather variables are likely, yet difficult to quantify. For example, cool temperatures and increased relative humidity can aid ice development, while cold, dry temperature increases sublimation of snow, thus reducing its insulating properties for ice (Murfitt et al. 2018). Although we hypothesized that increased winter precipitation would result in thinner ice (as reflected by earlier ice melt), the confidence intervals in the model suggest a possible unidentified interaction with the other predictors (Figueiras et al. 1998).

Counter to our hypothesis, we determined that shorter foraging periods (ice‐off to ice‐on) were associated with later emergence dates, although this variable was not represented in any of the top models. This trend could be due to various factors, unrelated to climate, or to unmeasured climate and ice condition variables. While not available for our research, body mass in the fall, influenced by resource quality pre‐immersion, can also influence timing of emergence. However, the trend can be species‐specific and quite variable even within species (Findlay‐Robinson et al. 2023). For yellow‐bellied marmot (Marmota flaviventer), emergence is earlier for heavier individuals, which allows individuals with higher energy reserves to reproduce earlier (Edic et al. 2020). Conversely, edible dormice with a higher body mass during hibernation emerged later than lighter individuals (Fietz et al. 2020). One consideration irrelevant to terrestrial species is the length of exposure of the pond to evapotranspiration, which in turn affects pond depth, especially in drier years. Shorter open water foraging periods might link to less evaporative water loss (hence deeper ponds) and, therefore, increased access to the food cache over winter. In our study, total precipitation from ice‐off to emergence in the top‐ranked models had a positive effect on emergence day. This scenario where shorter foraging periods increases time under the ice only exists if there is adequate insulation of the pond over winter to prevent thicker ice that would hinder access to the food cache. Unfortunately, we could not measure ice thickness or under‐ice water depth but expect these variables would provide additional insights into emergence dates for beavers. As noted by Aleksiuk (1970), beavers can expedite the timing of emergence by chewing through the ice, thus accounting for agency within the species that is independent of abiotic factors.

There are many studies examining emergence phenology for hibernating mammals (Wells et al. 2022; Findlay‐Robinson et al. 2023), yet studies on beavers have generally focused on their foraging behaviors following emergence (Aleksiuk 1970) or how spring green‐up phenology in plants influences survival, fitness, body mass, and reproductive success (Campbell et al. 2012, 2013). Although we did not document post‐emergence foraging, we noted anecdotal evidence of foraging behavior in low‐water years where the shoreline was exposed in fall, a condition that extended to the following spring. As a result, foraging distances increased by as much as 20 m due to receding shorelines and spring emergent and aquatic vegetation was scarce. To compensate for delayed growth of herbaceous vegetation, beavers cut aspens and willows in early spring more than expected from other studies (e.g., Hood and Bayley 2009), even stripping from felled trembling aspen trees large portions of the photosynthetic, chlorophyll‐containing living bark formed prior to leaf‐out in spring. In 2025, a very low water year, beavers cut adult aspen trees throughout the ice‐free period. Warmer temperatures play a role in photosynthesis in aspen bark (Bate and Canvin 1971), as seen in European beech ( Fagus sylvatica ) and silver birch ( Betula pendula ) (Wittmann and Pfanz 2007). Nutrients in poplar bark allow beavers to obtain high‐energy foods despite delayed growth of herbaceous plants due to drought conditions. Unlike hibernating mammals where phenological synchronicity with herbaceous plants can influence their reproductive success and overall survival (Macphie and Phillimore 2024), the ability of beavers to fell large trees, even prior to leaf‐out, helps mitigate occasional phenological mismatch and food scarcity in the spring. This adaptability is just one way in which beavers differ from hibernating mammals.

As noted, most research on emergence phenology is conducted on species that hibernate or experience daily torpor. Beavers represent a suite of species (freshwater semi‐aquatic mammals) that have other winter adaptations for aquatic habitats that are highly influenced by climate change, yet the opportunity to document emergence phenology over many years, as we have done in our study, is rare. As in our study, temperatures in northern latitudes are warming, especially in summer, and these changes will influence the duration of foraging periods, timing of ice formation and melt, and ultimately, survival of these species in their existing range. Little is known about how climate and ice phenology influence reproduction, fitness, and dispersal of beavers, yet they are a species with a disproportionate influence on the environment and other species dependent on freshwater habitats (Brazier et al. 2021). Given the increased presence of beavers in northern latitudes and their ongoing colonization of the Arctic tundra (Tape et al. 2022), further research is warranted on factors influencing emergence phenology of beavers, including differences across larger geographic areas, and implications for other species living at the interface between aquatic and terrestrial environments.

Author Contributions

Glynnis A. Hood: conceptualization, data collection, analysis, writing – original draft. D.L. (Dee) Patriquin: conceptualization, data collection, review and editing.

Disclosure

Permits: The current permit with Alberta Parks is 25‐019. Alberta Parks has granted permits for this research since January 2008.

Ethics Statement

The University of Alberta's Animal Care and Use Committee (ACUC) approved this research under protocol AUP00000073.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Mean average temperature (°C), mean minimum temperature (°C), and mean maximum temperature (°C) with standard deviation (SD), and total precipitation (mm) by beaver year (April 1—March 31) from 2007/2008 to 2024/2005 in Miquelon Lake Provincial Park, Alberta, Canada.

GCB-32-e70833-s001.docx (16.1KB, docx)

Acknowledgements

We thank Alberta Parks for their continued permitting of this research and the dozens of volunteers who participate in the authors' annual winter beaver lodge occupancy surveys at Miquelon Lake Provincial Park.

Data Availability Statement

Data are accessible through the Federated Research Data Repository (FRDR). https://doi.org/10.20383/103.01627.

References

  1. Aleksiuk, M. 1970. “The Seasonal Food Regime of Arctic Beavers.” Ecology 51: 264–270. 10.2307/1933662. [DOI] [Google Scholar]
  2. Aleksiuk, M. , and Cowan I. M.. 1969. “The Winter Metabolic Depression in Arctic Beavers ( Castor canadensis Kuhl) With Comparisons to California Beavers.” Canadian Journal of Zoology 47: 965–979. 10.1139/z89-094. [DOI] [Google Scholar]
  3. Baker, B. W. , and Hill E. P.. 2003. “Beaver ( Castor canadensis ).” In Wild Mammals of North America: Biology, Management, and Conservation, edited by Feldhamer G. A., Thompson B. C., and Chapman J. A., 2nd ed., 288–310. Johns Hopkins University Press. [Google Scholar]
  4. Bate, G. C. , and Canvin D. T.. 1971. “The Effect of Some Environmental Factors on the Growth of Young Aspen Trees ( Populus tremuloides ) in Controlled Environments.” Canadian Journal of Botany 49: 1443–1453. 10.1139/b71-203. [DOI] [Google Scholar]
  5. Bieber, C. , Turbill C., and Ruf T.. 2018. “Effects of Aging on Timing of Hibernation and Reproduction.” Scientific Reports 8: 13881. 10.1038/s41598-018-32311-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Brazier, R. E. , Puttock A., Graham H. A., Auster R. E., Davies K. H., and C. M. Brown . 2021. “Beaver: Nature's Ecosystem Engineers.” WIREs Water 8: e1494. 10.1002/wat2.1494. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bromley, C. K. , and Hood G. A.. 2013. “Beaver ( Castor canadensis ) Facilitate Early Access by Canada Geese ( Branta canadensis ) to Nesting Habitat and Open Water in Canada's Boreal Wetlands.” Mammalian Biology 78: 73–77. 10.1016/j.mambio.2012.02.009. [DOI] [Google Scholar]
  8. Burnham, K. P. , and Anderson D. R.. 2002. Model Selection and Multimodel Inference: A Practical Information‐Theoretical Approach. 2nd ed. Springer‐Verlag. [Google Scholar]
  9. Busher, P. E. 1996. “Food Caching Behavior of Beavers ( Castor canadensis ): Selection and Use of Woody Species.” American Midland Naturalist 135: 343–348. 10.2307/2426717. [DOI] [Google Scholar]
  10. Campbell, R. D. , Newman C., Macdonald D. W., and Rosell F.. 2013. “Proximate Weather Patterns and Spring Green‐Up Phenology Effect Eurasian Beaver ( Castor fiber ) Body Mass and Reproductive Success: The Implications of Climate Change and Topography.” Global Change Biology 19: 1311–1324. 10.1111/gcb.12114. [DOI] [PubMed] [Google Scholar]
  11. Campbell, R. D. , Nouvellet P., Newman C., Macdonald D. W., and Rosell F.. 2012. “The Influence of Mean Climate Trends and Climate Variance on Beaver Survival and Recruitment Dynamics.” Global Change Biology 18: 2730–2742. 10.1111/j.1365-2486.2012.02739.x. [DOI] [PubMed] [Google Scholar]
  12. Dickey, D. A. , and Fuller W. A.. 1979. “Distribution of the Estimators for Autoregressive Time Series With a Unit Root.” Journal of the American Statistical Association 74: 427–431. 10.1080/01621459.1979.10482531. [DOI] [Google Scholar]
  13. Dormann, C. F. , Elith J., Bacher S., et al. 2013. “Collinearity: A Review of Methods to Deal With It and a Simulation Study Evaluating Their Performance.” Ecography 36: 27–46. 10.1111/j.1600-0587.2012.07348.x. [DOI] [Google Scholar]
  14. Duguay, C. R. , Flato G. M., Jeffries M. O., Ménard P., Morris K., and Rouse W. R.. 2003. “Ice‐Cover Variability on Shallow Lakes at High Latitudes: Model Simulations and Observations.” Hydrological Processes 17: 3465–3483. 10.1002/hyp.1394. [DOI] [Google Scholar]
  15. Dyck, A. P. , and MacArthur R. A.. 1992. “Seasonal Patterns of Body Temperature and Activity in Free‐Ranging Beaver ( Castor canadensis ).” Canadian Journal of Zoology 70: 1668–1672. 10.1139/z92-232. [DOI] [Google Scholar]
  16. Edic, M. N. , Martin J. G., and Blumstein D. T.. 2020. “Heritable Variation in the Timing of Emergence From Hibernation.” Evolutionary Ecology 34: 763–776. 10.1007/s10682-020-10060-2. [DOI] [Google Scholar]
  17. Fietz, J. , Langer F., and Schlund W.. 2020. “They Like It Cold, but Only in Winter: Climate Mediated Effects on a Hibernator.” Functional Ecology 34: 2098–2109. 10.1111/1365-2435.13630. [DOI] [Google Scholar]
  18. Figueiras, A. , Domenech‐Massons J. M., and Cadarso C.. 1998. “Regression Models: Calculating the Confidence Interval of Effects in the Presence of Interactions.” Statistics in Medicine 17: 2099–2105. 10.1002/(SICI)1097-0258(19980930)17:18<2099::AID-SIM905>3.0.CO;2-6. [DOI] [PubMed] [Google Scholar]
  19. Findlay‐Robinson, R. , Deecke V. B., Weatherall A., and Hill D. L.. 2023. “Effects of Climate Change on Life‐History Traits in Hibernating Mammals.” Mammal Review 53: 84–98. 10.1111/mam.12308. [DOI] [Google Scholar]
  20. Fujisaki‐Manome, A. , Anderson E. J., Kessler J. A., Chu P. Y., Wang J., and Gronewold A. D.. 2020. “Simulating Impacts of Precipitation on Ice Cover and Surface Water Temperature Across Large Lakes.” Journal of Geophysical Research. Oceans 125: e2019JC015950. 10.1029/2019JC015950. [DOI] [Google Scholar]
  21. Geiser, F. 2020. “Seasonal Expression of Avian and Mammalian Daily Torpor and Hibernation: Not a Simple Summer‐Winter Affair.” Frontiers in Physiology 11: 436. 10.3389/fphys.2020.00436. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Harrington, R. , Woiwod I., and Sparks T.. 1999. “Climate Change and Trophic Interactions.” Trends in Ecology & Evolution 14: 146–150. 10.1016/S0169-5347(99)01604-3. [DOI] [PubMed] [Google Scholar]
  23. He, Z. , and Pomeroy J. W.. 2023. “Assessing Hydrological Sensitivity to Future Climate Change Over the Canadian Southern Boreal Forest.” Journal of Hydrology 624: 129897. 10.1016/j.jhydrol.2023.129897. [DOI] [Google Scholar]
  24. Hood, G. A. 2020a. Semi‐Aquatic Mammals: Ecology and Biology. Johns Hopkins University Press. [Google Scholar]
  25. Hood, G. A. 2020b. “Not All Ponds Are Created Equal: Long‐Term Beaver ( Castor canadensis ) Lodge Occupancy in a Heterogeneous Landscape.” Canadian Journal of Zoology 98: 210–218. 10.1139/cjz-2019-0066. [DOI] [Google Scholar]
  26. Hood, G. A. , and Bayley S. E.. 2009. “A Comparison of Riparian Plant Community Response to Herbivory by Beaver ( Castor canadensis ) and Ungulates in Canada's Boreal Mixed‐Wood Forest.” Forest Ecology and Management 258: 1979–1989. 10.1016/j.foreco.2009.07.052. [DOI] [Google Scholar]
  27. Hufnagl, S. , Franceschini‐Zink C., and Millesi E.. 2011. “Seasonal Constraints and Reproductive Performance in Female Common Hamsters Cricetus cricetus .” Mammalian Biology 76: 124–128. 10.1016/j.mambio.2010.07.004. [DOI] [PubMed] [Google Scholar]
  28. Jenkins, S. H. , and Busher P. E.. 1979. “ Castor canadensis .” Mammalian Species 120: 1–8. 10.2307/3503787. [DOI] [Google Scholar]
  29. Kendall, M. G. 1975. Rank Correlation Methods. 4th ed. Charles Griffin. [Google Scholar]
  30. Lai, Y. , and Dzombak D. A.. 2020. “Use of the Autoregressive Integrated Moving Average (ARIMA) Model to Forecast Near‐Term Regional Temperature and Precipitation.” Weather and Forecasting 35: 959–976. 10.1175/WAF-D-19-0158.1. [DOI] [Google Scholar]
  31. Lancia, R. A. , Dodge W. E., and Larson J. S.. 1982. “Winter Activity Patterns of Two Radio‐Marked Beaver Colonies.” Journal of Mammalogy 63: 598–606. 10.2307/1380264. [DOI] [Google Scholar]
  32. Lorentz, B. M. , and Hood G. A.. 2025. “Drivers of Winter Habitat Selection by Muskrats in Southern Boreal Wetlands of Alberta, Canada.” Mammalian Biology 105: 201–213. 10.1007/s42991-024-00469-5. [DOI] [Google Scholar]
  33. Macphie, K. H. , and Phillimore A. B.. 2024. “Phenology.” Current Biology 34: R183–R188. 10.1016/j.cub.2024.01.007. [DOI] [PubMed] [Google Scholar]
  34. Mann, H. B. 1945. “Nonparametric Tests Against Trend.” Econometrica 13: 245–259. 10.2307/1907187. [DOI] [Google Scholar]
  35. Mayer, M. , Künzel F., Zedrosser A., and Rosell F.. 2017. “The 7‐Year Itch: Non‐Adaptive Mate Change in the Eurasian Beaver.” Behavioral Ecology and Sociobiology 71: 32. 10.1007/s00265-016-2259-z. [DOI] [Google Scholar]
  36. Mayer, M. , Zedrosser A., and Rosell F.. 2017. “When to Leave: The Timing of Natal Dispersal in a Large, Monogamous Rodent, the Eurasian Beaver.” Animal Behaviour 123: 375–382. 10.1016/j.anbehav.2016.11.020. [DOI] [Google Scholar]
  37. McCain, C. M. , and King S. R.. 2014. “Body Size and Activity Times Mediate Mammalian Responses to Climate Change.” Global Change Biology 20: 1760–1769. 10.1111/gcb.12499. [DOI] [PubMed] [Google Scholar]
  38. Miller, L. K. 1967. “Microclimate of the Northern Beaver: A Constructed Habitat.” Biometeorology 3: 288 As cited in Baker, B. W., and E. P. Hill. 2003. Beaver (Castor canadensis). In Wild Mammals of North America: Biology, Management, and Conservation (2nd ed., pp. 288–310), edited by G. A. Feldhamer, B. C. Thompson, and J. A. Chapman. Johns Hopkins University Press. [Google Scholar]
  39. Milligan, H. E. , and Humphries M. M.. 2010. “The Importance of Aquatic Vegetation in Beaver Diets and the Seasonal and Habitat Specificity of Aquatic‐Terrestrial Ecosystem Linkages in a Subarctic Environment.” Oikos 119: 1877–1886. 10.1111/j.1600-0706.2010.18160.x. [DOI] [Google Scholar]
  40. Murfitt, J. C. , Brown L. C., and Howell S. E.. 2018. “Estimating Lake Ice Thickness in Central Ontario.” PLoS One 13: e0208519. 10.1371/journal.pone.0208519. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Prather, R. M. , Dalton R. M., Barr B., et al. 2023. “Current and Lagged Climate Affects Phenology Across Diverse Taxonomic Groups.” Proceedings of the Royal Society B 290, no. 1990: 20222181. 10.1098/rspb.2022.2181. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Price, D. T. , Alfaro R. I., Brown K. J., et al. 2013. “Anticipating the Consequences of Climate Change for Canada's Boreal Forest Ecosystems.” Environmental Reviews 21: 322–365. 10.1139/er-2013-0042. [DOI] [Google Scholar]
  43. R Core Team . 2025. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. https://www.R‐project.org/. [Google Scholar]
  44. Reynolds, P. S. 1992. “Behavioural and Biophysical Ecology of Beaver Castor canadensis in North‐Central Wisconsin.” PhD Dissertation, University of Wisconsin‐Madison. https://www.proquest.com/openview/4efadcfdc3a9c77cb0ae29bde7176703/1?pq‐origsite=gscholar&cbl=18750&diss=y.
  45. Singh, H. , Pirani F. J., and Najafi M. R.. 2020. “Characterizing the Temperature and Precipitation Covariability Over Canada.” Theoretical and Applied Climatology 139: 1543–1558. 10.1007/s00704-019-03062-w. [DOI] [Google Scholar]
  46. Smith, D. W. , Peterson R. O., Drummer T. D., and Sheputis D. S.. 1991. “Over‐Winter Activity and Body Temperature Patterns in Northern Beavers.” Canadian Journal of Zoology 69: 2178–2182. 10.1139/z91-304. [DOI] [Google Scholar]
  47. Stephenson, A. B. 1969. “Temperatures Within a Beaver Lodge in Winter.” Journal of Mammalogy 50: 134–136. 10.2307/1378645. [DOI] [Google Scholar]
  48. Sun, L. , Müller‐Schwarze D., and Schulte B. A.. 2000. “Dispersal Pattern and Effective Population Size of the Beaver.” Canadian Journal of Zoology 78: 333–513. 10.1139/z99-226. [DOI] [Google Scholar]
  49. Tape, K. D. , Clark J. A., Jones B. M., et al. 2022. “Expanding Beaver Pond Distribution in Arctic Alaska, 1949 to 2019.” Scientific Reports 12: 7123. 10.1038/s41598-022-09330-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Tape, K. D. , Jones B. M., Arp C. D., Nitze I., and Grosse G.. 2018. “Tundra be Dammed: Beaver Colonization of the Arctic.” Global Change Biology 24: 4478–4488. 10.1111/gcb.14332. [DOI] [PubMed] [Google Scholar]
  51. Turbill, C. , and Prior S.. 2016. “Thermal Climate‐Linked Variation in Annual Survival Rate of Hibernating Rodents: Shorter Winter Dormancy and Lower Survival in Warmer Climates.” Functional Ecology 30: 1366–1372. 10.1111/1365-2435.12620. [DOI] [Google Scholar]
  52. Wells, C. P. , Barbier R., Nelson S., Kanaziz R., and Aubry L. M.. 2022. “Life History Consequences of Climate Change in Hibernating Mammals: A Review.” Ecography 2022, no. 6: e06056. 10.1111/ecog.06056. [DOI] [Google Scholar]
  53. Wittmann, C. , and Pfanz H.. 2007. “Temperature Dependency of Bark Photosynthesis in Beech ( Fagus sylvatica L.) and Birch ( Betula pendula Roth.) Trees.” Journal of Experimental Botany 58: 4293–4306. 10.1093/jxb/erm313. [DOI] [PubMed] [Google Scholar]
  54. Woo, M. K. , Thorne R., Szeto K., and Yang D.. 2008. “Streamflow Hydrology in the Boreal Region Under the Influences of Climate and Human Interference.” Philosophical Transactions of the Royal Society, B: Biological Sciences 363: 2251–2260. 10.1098/rstb.2007.2197. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Zhang, Z. , Bortolotti L. E., Li Z., Armstrong L. M., Bell T. W., and Li Y.. 2021. “Heterogeneous Changes to Wetlands in the Canadian Prairies Under Future Climate.” Water Resources Research 57: e2020WR028727. 10.1029/2020WR028727. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Table S1: Mean average temperature (°C), mean minimum temperature (°C), and mean maximum temperature (°C) with standard deviation (SD), and total precipitation (mm) by beaver year (April 1—March 31) from 2007/2008 to 2024/2005 in Miquelon Lake Provincial Park, Alberta, Canada.

GCB-32-e70833-s001.docx (16.1KB, docx)

Data Availability Statement

Data are accessible through the Federated Research Data Repository (FRDR). https://doi.org/10.20383/103.01627.

Articles from Global Change Biology are provided here courtesy of Wiley

RESOURCES