Moving for Two: How Pregnancy Shapes Moose Movement in a Mined Landscape
Purpose of the study
Human driven landscape change is increasingly altering wildlife movement, habitat selection and predator prey dynamics in boreal ecosystems. In northern Alberta, industrial activities associated with oil sands development have substantially altered habitat structure through land clearing, infrastructure expansion and increased road networks. These changes influence spatial behaviour in large herbivores such as mooose (Alces alces) and their primary predator, the gray wolf (Canis lupus), by modifying movement patterns across the landscape.
Previous research in the Athabasca oil sands region has demonstrated that human disturbance can both elevate and reduce predation risk for moose. As a result, moose in industrially modified landscapes have to navigate the spatial trade off between resource acquisition and risk avoidance.
Figure 1. Adult female moose in wetland habitat . Source: (Strzelecki,. J 2001)
This study examines the spatial ecology of 25 GPS collared adult female moose (Fig 1) monitored in a heterogenous landscape. There are currently gaps in research comparing fine scale movement behaviour and space use between pregnant and non-pregnant females during late gestation and the calving period. Reproductive status is expected to alter energetic demands and predation related constraints. The analysis of this study focuses on the home ranges, resource use and movement behaviour of moose during the calving season to help understand how calving influences spatial behaviour to aid in effective management and conservation.
Context
Importance of Moose
Moose are the largest members of the deer family and play an important role in shaping boreal ecosystems (Figure 2). As large herbivores, they influence vegetation composition, nutrient cycling and plant establishment through selective foraging and trampling (Rodgers 2001). Therefore, their movement patterns and population dynamics have ecological effects across northern landscapes. Moose populations are increasingly exposed to multiple environmental pressures, including habitat degradation, climate change, predation and disease. Rising temperatures are expected to increase thermal stress and parasite loads, resulting in decreased survival and reproductive success (Elzinga, Beckford and Strickland 2023). Movement decisions can be interpreted through ecological frameworks such as optimal foraging theory, in which individuals balance energy acquisition against risks such as predation and human disturbance. These trade offs are especially pronounced in modified landscapes.
Figure 2. Two adult female moose in a dense forest habitat. Source: (Barette,.J 2009).
Reproductive Status and Movement Behaviour
In rapidly changing environments, key ecological processes such as reproduction, survival and movement are being altered (Hansson et al. 2014). Reproductive status is a key driver of spatial behaviour in female moose. This study focuses exclusively on adult females, and will be comparing the spatial ecology of pregnant and non pregnant moose. The rutting period typically occurs between early September to November and calving usually takes place between May and June (Naughton and Canadian Museum of Nature. 2012). Females with calves are expected to prioritise predator avoidance, potentially resulting in reduced movement and more constrained home ranges. Moose are largely solitary animals, however, those in calf tend to remain in closer proximity potentially exhibiting shared home ranges.
Study Area: Alberta, Canada
The study was conducted in Alberta, Canada, a region characterised by heterogeneous landscapes including boreal forest, wetlands and rugged terrain (Figure 3). This diverse environment provides a range of habitats that influence moose distribution and movement patterns. The climate of the region is highly seasonal, with extremely low temperatures and heavy snowfall during winter and relatively warm summers. Climate projections indicate that temperatures are expected to increase by approximately 2–8°C by the end of the 21st century (Barber and Sesser 2019). These changes are likely to alter habitat suitability, forage availability and movement behaviour. Long term monitoring is therefore critical for understanding how moose respond to these environmental changes.
In addition to natural factors, anthropogenic disturbance plays a major role in shaping the landscape. Roads, industrial infrastructure and land use change associated with oil sands development have fragmented habitats and altered movement corridors. The study area lies within an extensive oil sands region the Athabasca River, approximately 20km north of Fort McMurray (Neilson and Boutin 2017). Oil sands extraction involves large scale surface mining, resulting in habitat loss, air pollution, excessive water usage and toxic waste creation (Johnson et al., 2008).
Figure 3. Interactive map showing GPS telemetry locations of female moose in the Alberta study area and in context to location within Canada. Each colour represents an individual moose.
Importance of Spatial Ecology and GPS telemetry
Spatial ecology provides a framework for understanding how species interact with their environment across space and time. Space use, habitat selection and movement behaviour determine how animals access resources, avoid risks and interact with other species across landscapes (Fletcher and Fortin 2019). These processes are particularly important for wide ranging species such as moose that experience natural and anthropogenic pressures. When looking at spatial ecology it is important to integrate movement, habitat selection and space use together to provide a more realistic framework (Van Moorter et al. 2016).
Advances in GPS telemetry have benefitted the study of spatial ecology by enabling the collection of high resolution location data across large spatial scales. Platforms such as Movebank allow easy data sharing of large scale, multi species comparitive analysis (Kays et al. 2015). However, GPS telemetry also presents challenges, including potential animal disturbance from collars and data gaps resulting from signal obstruction in dense vegetation or rugged terrain (D’eon and Delparte 2005).
Analytical Approach
GPS Telemetry Data Cleaning
The dataset was obtained from movebank (Boutin et al. 2014) and contained location data from 25 adult female moose monitored between 2010-2012. The dataset had 133,518 GPS locations with fixes recorded at every 3 hours with significant differences in the length of tracking days between each moose (Table A1-Appendix). For reproductive status analysis March-June 2010 was selected as it encompasses late gestation (March-April), pre calving behaviour (May) and early calving period (late May-June). During this period 22466 locations were available for analysis.
Individuals with insufficient data were removed including yl19 as the individual died after only 60 days and yl21 was also removed as calving status was unknown. Data were screened for missing values, duplicate timestamps and exploratory analyis was performed, including average speed (mean 44.29m/h, median 16.4m/h) and mean step length (mean 135.49m, median 49.58m).
Home Range Analysis
Minimum Convex Polygon (95% MCP) were used to describe overall spatial extent of moose range across the full three year monitoring period. MCPs represent the smallest possible polygon enclosing 95% of observed locations and are widely used due to their simplicity. However, MCPs have known limitations including sensitivity to sample size and overestimation of space use (Nilsen, Pedersen and Linnell 2008).
To address these limitations, 95% autocorrelated kernel density estimation (AKDE) was applied to quantify home range size during the spring calving period. AKDE estimates the area containing 95% of predicted space use while accounting for autocorrelation in movement data. Unlike conventional KDE, AKDE incorporates continuous time movement processes, allowing more accurate and less biased estimates (Fleming et al. 2015). The combined approach of MCP and AKDE allows for both broad and detailed insights into moose spatial ecology.
Resource Selection Analysis
Habitat loss is recognised as the main driver of biodiversity decline and resource availability plays a key role in determining species abundance (Fischer and Lindenmayer 2007). Resource use was quantified using step selection functions (SSF) by comparing environmental characteristics at observed movement steps (“used” locations) with characterisitics at available unused locations. This approach accounts for movement constraints and spatial autocorrelation.
SSF analysis was also conducted during the calving period with covariates including land cover, ruggedness and reproductive status, while movement was characterised using step length and turning angle. Models were fitted with a Bayesian framework using integrated nested laplace approximation (INLA), allowing incorporation of prior knowledge for parameter estimates. Reproductive status was included as a covariate to assess whether females with calves exhibited systematic differences in step selection behaviour after accounting for habitat, terrain and movement constraints.
Movement Analysis
Animal movement provides insight into how individuals respond to environmental and biological processes and can be used to understand behavioural changes (Kays et al. 2015). Movement behaviour during the calving period was examined using hidden Markov models (HMMs), which identify underlying behavioural states from observed movement metrics. This is particularly useful in ecology as it distinguishes based on step length (distance between successive locations) and turning angle distribution (change in direction between successive steps) (Bouguila, Fan and Amayri 2022).
A two state HMM framework was used, representing restricted (foraging or resting) and travelling movement. Models were fitted with the momentuHMM package in R, which provides a flexible framework for analysing animal movement data and integrating biological covariates (McClintock, Michelot and Goslee 2018). This approach allowed assessment of how reproductive status influenced both movement behaviour and the likelihood of switching between behavioural states during the calving period.
Outcomes
Home Ranges
Minimum Convex Polygon (Full study period: 2010-2012)
The 95% MCP estimates revealed substantial variation in home range size and spatial distribution among individual moose (Figure 4). Home ranges were widely dispersed across the study area, with several individuals (e.g. yl1, yl7,yl13 and yl20) exhibiting larger MCPs, indicating greater spatial use while smaller MCPs could indicate more restricted ranges, differences in sampling duration or collar performance.
MCPs were often irregular, suggesting non-uniform space use rather than consistent utilisation of the area. Spatial clustering in central regions with overlap between multiple individuals was observed suggesting shared use of key habitat features or resources. Comparatively, some individuals occupied more isolated ranges with minimal overlap, highlighting potential differences in habitat preferences or movement behaviour.
Figure 4. 95% MCP home range estimates for individual moose based on GPS telemetry data collected over the three year period in Alberta. Each polygon represents an individuals spatial extent with unique colours corresponding to different individuals.
Estimated 95% MCP home range sizes ranged from 175.57 ha (yl19) to 12,963.99 ha (yl13), showing substantial inter individual variation in space use across the study period (Table 1). The mean home range size was 5,335.41 ha (± 3,368.28 SD), with a median of 4,842.65 ha, suggesting a moderately right skewed distribution driven by a small number of individuals with particularly large home ranges. The broad range and high variability indicate substantial heterogeneity in spatial ecology among individuals.
Table 1. Summary 95% MCP home range sizes (ha) for each individual moose, with data spanning the full three year data set. Statistics (mean, median, standard deviation , minimum and maximum values) highlight comparative variation amongst data.
AKDE Home Ranges During Calving Period (Spring 2010)
95% AKDE arevealed variation and overlap in home ranges during the calving period, regardless of reproductive status (Figure 5). Although moose in calf were further subdivided according to the number of calves present, uneven sample sizes limited robust statistical comparison among these groups.
Figure 5. 95% AKDE home range estimates of non-pregnant (blue), pregnant with 1 calf (orange) and pregnant with 2 calves (red) during the spring calving period for visual representation.
Pregnant moose had home range sizes ranging from approximately 12 to 173km², while non pregnant ranged from 13 to 133 km². Mean home range size was similar between the two groups, with moose in calf averaging 62.41 km² and moose not in calf averaging 64.80 km². Confidence intervals overlapped extensively among individuals and between reproductive groups, indicating high individual variability.
Table 2. Individual level (Spring 2010) home range estimates using 95% AKDE. Moose with no calf and ≥ 1 calf separated for statistical comparison. The means of both groups were included for total group comparison.
Resource Selection
Step selection during the calving period varied strongly among habitat types. Moose showed positive selection for developed or industrial areas (mean =0.553, 95% CI: 0.19 to 0.91), forage habitats (mean =0.225, 95% CI: 0.16 to 0.30), wetlands (mean = 0.172, 95% CI: 0.096 to 0.247) and woodland (mean =0.125, 95% CI: 0.063 to 0.187) relative to the mean available habitat (Figure 6) . In contrast, open water was strongly avoided (mean =−1.075, 95% CI:−1.643 to −0.506).
The presence of a calf had no detectable effect on step selection (mean = −0.001, 95% CI: −0.085 to 0.083), indicating broadly similar habitat use. Terrain ruggedness likewise showed no clear influence on step selection (mean =0.001, 95% CI:−0.035 to 0.037). Together, these results indicate that moose primarily differ in their selection among habitat types rather than in reproductive state specific habitat use.
Figure 6. Bayesian SSF coefficients for moose during the calving period. Points represent posterior mean estimates and bars indicate 95% credible intervals. Coefficients are shown on the log‑odds scale and habitat effects are centered around the mean available habitat. Effect types include habitat (green), Reproductive status (blue) and terrain (brown).
Movement patterns and Behaviour
Step length distribution during the spring calving period were strongly right skewed for both reproductive groups (Figure 7). Most movement steps were short, with frequency declining rapidly as step length increased, although long step lengths exceeding 500m occurred in both groups. Pregnant moose exhibited a higher concentration of short step lengths compared to non pregnant, indicating more restricted movement during the calving period. In contrast, moose not in calf showed a broader distribution of step lengths and higher frequency of long movements.
Figure 7. Distribution of step lengths for moose that were not pregnant (left) and pregnant (right) during spring calving period. Histograms represent frequency (y axis) and step length (≤ 1500m, x axis).
Turning angle distributions during the spring calving period were relatively similar between reproductive groups, with values broadly spanning the full range of turning angles (Figure 8). Pregnant moose showed a higher concentration of turning angles near zero, suggesting greater directional persistence. Whereas, non pregnant moose had a flatter distribution indicative of more variable movement direction. Both step length and turning angle graphs are only a visual interpretation.
Figure 8. Distribution of turning angles for non pregnant moose (left) and pregnant (right) during the spring calving period. Histograms represent the density (y axis) and turning angles (radians, x axis).
Interpretation of main findings
Effect of reproductive status on Moose movement
Based on previous studies of ungulate reproductive ecology, it was expected that pregnant moose would exhibit smaller home ranges and more localised, clustered movement as a strategy to reduce energetic expenditure and predation risk during the calving period (Bonar et al. 2018). However, the AKDE results showed substantial overlap in home range size between reproductive groups, with several moose in calf exhibiting home ranges comparable to or larger than those of moose not in calf (Fig 5). The individual with the largest observed home range size (yl13) was pregnant, illustrating that large home range extent is not necessarily consistent with reproductive status (Table 2).
Additionally the SSF analysis showed calf presence (pregnant or birthed calf) had no detectable effect on SSF after accounting for habitat type, terrain ruggedness and movement constraints. These combined results show that home range and resource selection showed minimal difference between reproductive groups. Seasonal shifts in forage availability likely contribute to this pattern as both moose in calf and not in calf will be using the same resources available. During spring, moose diets transition from woody browse to emerging vegetation and aquatic plants, which are patchily distributed across wetlands, forests and foraged areas (Tischler et al. 2019). The SSF results reflected this ecological context, with positive selection observed for these habitats, which provide a combination of nutritional resources, concealment and structural cover (Figure 6).
Visual inspection of MCP map (Fig 4), indicated that the home range of yl13 encompassed the edge of a large body of water, consistent with use of aquatic vegetation. Contrastingly, yl25 a pregnant moose with two calves had the smallest home range, also centred on wetland environments. Together, these patterns suggest that larger home range size does not necessarily indicate increased mobility but may reflect the spatial distribution of essential resources. While overall home range extent were similar between reproductive groups, pronounced differences emerged in fine scale movement behaviour.
Analysis of step length and turning angle distributions showed that non pregnant moose exhibited a wider range of step lengths, including more long distance movements and greater directional variability. HMM model reinforced these findings, demonstrating that moose in calf were less likely to transition from restricted movement to travelling states and spent a greater proportion of time in restricted movement. This indicates that pregnant moose constrained their movement behaviour within their home ranges, likely balancing energetic efficiency with offspring protection during late gestation and early calving. These results highlight the importance of distinguishing between home range extent and fine scale movement behaviour when evaluating ecological responses to reproductive state.
Influence of a Mined Landscape
Fig 11. Visual extent of the disruption to the land caused by oils sands mining, Alberta, Canada.
Human modified landscapes can substantially alter predator prey dynamics and movement corridors. In boreal ecosystems, populations are often more resilient to environmental change when individuals exhibit flexible and diverse patterns of resource use (Banks et al., 2013). The oil sands region of northern Alberta represents a highly altered landscape where such flexibility may be increasingly important.
One of the biggest threats to moose is predation, Gray wolves are the primary cause of juvenile mortality (Severud et.al, 2015). Previous studies in the Athabasca oil sands region have shown that mining activity alters wolf movement by removing natural vegetation and creating hard linear edges, which concentrate predator movement along movement corridors and increase encounter rates with moose (Neilson & Boutin, 2017). These studies suggest elevated predation risk near mine perimeters, particularly for vulnerable juveniles. Other research has demonstrated that moose often avoid roads despite the presence of high quality forage, indicating a trade off between resource availability and perceived risk (Brown et al., 2018).
In contrast, wolves have been found to avoid areas with high human activity near mining facilities, whereas moose do not show comparable avoidance (Neilson & Boutin, 2016). This may reduce predator presence in some industrial areas, potentially creating localized refuges for moose. The SSF results are consistent with this evidence, showing positive selection for developed or industrial areas during the calving period. Developed habitats in this study were classified using Human Footprint Inventory data and included mining areas, roads and settlements.
The unexpected similarity in home range size between reproductive groups may therefore reflect behavioural responses to a complex landscape shaped by industrial disturbance. Pregnant moose may need to extend their home ranges to avoid areas of high predator activity or to exploit areas of lower risk associated with human presence. Alternatively, non pregnant moose may reduce their spatial extent for similar reasons. These findings highlight the importance of considering human modified landscapes when interpreting animal movement and space use patterns.
Limitations
Several limitations should be considered when interpreting these results. Firstly, substantial individual variability was observed within both reproductive groups, which may reflect differences in the timing of pregnancy and calving within the March–June analysis window. This period encompasses late gestation, parturition, and early calf dependence, during which behavioural strategies may change rapidly. Home range estimates integrate movement across this entire period and may therefore mask movement that occur immediately before or after calving.
Secondly, sample sizes were uneven between reproductive groups, limiting statistical power to detect subtle differences. Future studies would benefit from larger sample sizes and balanced group numbers. Thirdly, GPS telemetry data are subject to positional error, particularly in dense forest cover and rugged terrain, which may introduce uncertainty into fine scale movement estimates.
Conservation Perspectives
Study findings to support management
The main conclusions of this study demonstrate how industrial land use can alter the natural movement behaviour of moose. Similar responses have been documented in other ungulate species, such as caribou, which exhibit elevated stress levels and poor nutrition in high human activity areas (Wasser et al. 2011). These responses may generate cascading effects, as wolves would preferentially predate caribou over moose in disturbed areas accelerating caribou decline or wolf numbers may decline as the mined land is providing refugia for moose (Serrouya et al. 2017).
Strong selection for forested habitats and wetland habitats by moose indicate that these environments provide essential resources for forage and cover. Conservation strategies should therefore prioritise these areas, particularly in areas used repeatedly during the spring calving period. Preserving these habitats is especially important for breeding females providing necessary cover and extra forage required.
Road networks represent a particularly important management concern with moose (Figure 12). Increased road density has been linked to higher rates of moose vehicle collisions, resulting in increased human safety risks and elevated moose mortality (Brubacher et al. 2016). Vegetation management along road verges may further enhance these effects, as early successional growth following clearing can attract browsing moose, particularly during spring when forage demand is high (Tanner and Leroux 2015). Long term management strategies that retain mature vegetation near road corridors, rather than promoting young vegetation, may reduce roadside foraging and collision risk while improved protection for moose.
Figure 12. Female adult moose using a road as movement corridor. Source: (Försäkring,. R 2016)
Future monitoring and adaptive management
Continued monitoring of moose populations is essential for evaluating how ongoing habitat modification, climate variability and management interventions influence space use and movement behaviour over time. While GPS telemetry provides detailed insights into movement and habitat selection, it is not entirely non-invasive and can be logistically and financially demanding. Future monitoring efforts could benefit from incorporating complementary methods such as non invasive genetic sampling using saliva from browsed twigs or faecal DNA. These approaches have proven effective for estimating population size and structure through genetic mark recapture with reduced animal handling (Jansson, Giles and Spong 2025). These monitoring techniques would support adaptive management under ongoing habitat modification and climate change by enabling adjustments to land use planning and conservation strategies.
References
Banks, S.C., Lindenmayer, D.B., Wood, J.T., McBurney, L., Blair, D. and Blyton, M.D.J., 2013. Can Individual and Social Patterns of Resource Use Buffer Animal Populations against Resource Decline? PloS One [online], 8 (1), pp.e53672. DOI: 10.1371/journal.pone.0053672 .
Barber, V., and Sesser, A.L., 2019. Drivers of landscape change in the Northwest Boreal Region [eBook]. Fairbanks, AK: University of Alaska Press.
Barette,.J(2009). Moose, female, in Jacques-Cartier River, Jacques-Cartier National Park, Quebec, Canada. Wikimedia Commons. https://commons.wikimedia.org/wiki/File:Alces_alces_PJC3.jpg (CC BY-SA 3.0)
Bonar, M., Ellington, E.H., Lewis, K.P. and Vander Wal, E., 2018. Implementing a novel movement-based approach to inferring parturition and neonate caribou calf survival. PloS One [online], 13 (2), pp.e0192204. DOI: 10.1371/journal.pone.0192204 .
Bouguila, N., Fan, W. and Amayri, M., 2022. Hidden Markov Models and Applications [eBook]. 1st ed. Cham: Springer International Publishing.
Boutin, S., Neilson, E.W. and de la Mare, C, Bohm, H., (2014) Wildlife habitat effectiveness and connectivity: Moose ecology project summary report 2010–2012. University of Alberta. Movebank Data Repository, Study ID: 302664172. Available at: https://www.movebank.org (Accessed: [15 March 2026])
Brown, C.L., Kielland, K., Brinkman, T.J., Gilbert, S.L. and Euskirchen, E.S., 2018. Resource selection and movement of male moose in response to varying levels of off‐road vehicle access. Ecosphere (Washington, D.C) [online], 9 (9), pp.n/a. DOI: 10.1002/ecs2.2405 .
Brubacher, J.R., Chan, H., Erdelyi, S., Schuurman, N. and Amram, O., 2016. The Association between Regional Environmental Factors and Road Trauma Rates: A Geospatial Analysis of 10 Years of Road Traffic Crashes in British Columbia, Canada. PloS One [online], 11 (4), pp.e0153742. DOI: 10.1371/journal.pone.0153742 .
D’eon, R.G., and Delparte, D., 2005. Effects of radio-collar position and orientation of GPS radio-collar performance, and the implications of PDOP in data screening. The Journal of Applied Ecology [online], 42 (2), pp.383–388. DOI: 10.1111/j.1365-2664.2005.01010.x .
Elzinga, D.C., Beckford, C. and Strickland, W.C., 2023. A mathematical model of the impacts of climate change on the winter tick epizootic in moose. Ecological Modelling [online], 483, pp.110421. DOI: 10.1016/j.ecolmodel.2023.110421 .
Fischer, J., and Lindenmayer, D.B., 2007. Landscape modification and habitat fragmentation: a synthesis. Global Ecology and Biogeography [online], 16 (3), pp.265–280. DOI: 10.1111/j.1466-8238.2007.00287.x .
Fleming, C.H., Fagan, W.F., Mueller, T., Olson, K.A., Leimgruber, P. and Calabrese, J.M., 2015. Rigorous home range estimation with movement data: a new autocorrelated kernel density estimator. Ecology (Durham) [online], 96 (5), pp.1182–1188. DOI: 10.1890/14-2010.1 .
Fletcher, R., and Fortin, M., 2019. Spatial Ecology and Conservation Modeling: Applications with R [eBook]. 1st ed. Cham: Springer Nature.
Försäkring,. R (2016). A cow moose in Sweden. Wikimedia Commons.https://commons.wikimedia.org/wiki/File:Moose_on_the_road_in_Sweden_001.jpg (CC BY-SA 3.0)
Hansson, L., êAkesson, S. and Åkesson, S., 2014. Animal movement across scales [eBook]. Oxford: Oxford University Press.
Hu, S., Jiang, G., Ma, J., Zhang, M. and Liu, H., 2013. Resource selection of moose (Alces alces cameloides) and their response to human disturbances in the northwestern slope of Lesser Khingan Mountains, northeastern China. Journal of Forestry Research [online], 24 (4), pp.727–734. DOI: 10.1007/s11676-013-0411-4 .
Jansson, J.L., Giles, B. and Spong, G., 2025. Estimating the population size of a semi-isolated moose (Alces alces) population from two sources of non-invasively collected DNA. European Journal of Wildlife Research [online], 71 (5), pp.100. DOI: 10.1007/s10344-025-01982-9 .
Johnson, E.A., Miyanishi, K., Ostfeld, R.S. and Schlesinger, W.H., 2008. Creating New Landscapes and Ecosystems: The Alberta Oil Sands. Annals of the New York Academy of Sciences [online], 1134 (1), pp.120–145. DOI: 10.1196/annals.1439.007 .
Kays, R., Crofoot, M.C., Jetz, W. and Wikelski, M., 2015. Terrestrial animal tracking as an eye on life and planet. Science (American Association for the Advancement of Science) [online], 348 (6240), pp.1222. DOI: 10.1126/science.aaa2478 .
McClintock, B.T., Michelot, T. and Goslee, S., 2018. momentuHMM: R package for generalized hidden Markov models of animal movement. Methods in Ecology and Evolution [online], 9 (6), pp.1518–1530. DOI: 10.1111/2041-210X.12995 .
Naughton, D., and Canadian Museum of Nature., 2012. The natural history of Canadian mammals [eBook]. Toronto Ont.] ;: University of Toronto Press.
Neilson, E., and Boutin, S., 2016. Moose refugia from predation by wolves near mines in the Athabasca oil sands. PeerJ Preprints [online], . DOI: 10.7287/peerj.preprints.1967v1 .
Neilson, E.W., and Boutin, S., 2017. Human disturbance alters the predation rate of moose in the Athabasca oil sands. Ecosphere (Washington, D.C) [online], 8 (8), pp.n/a. DOI: 10.1002/ecs2.1913 .
Nilsen, E.B., Pedersen, S. and Linnell, J.D.C., 2008. Can minimum convex polygon home ranges be used to draw biologically meaningful conclusions. Ecological Research [online], 23 (3), pp.635–639. DOI: 10.1007/s11284-007-0421-9 .
Rai, V., 2013. Spatial ecology [eBook]. 1st ed. Sharjah, U.A.E: Bentham Science Publishers.
Rodgers, A., 2001. Moose [eBook]. Grantown-on-Spey: Colin Baxter Photography.
RUCKSTUHL, K.E., and NEUHAUS, P., 2002. Sexual segregation in ungulates: a comparative test of three hypotheses. Biological Reviews of the Cambridge Philosophical Society [online], 77 (1), pp.77–96. DOI: 10.1017/S1464793101005814 .
Serrouya, R., McLellan, B.N., van Oort, H., Mowat, G. and Boutin, S., 2017. Experimental moose reduction lowers wolf density and stops decline of endangered caribou. PeerJ (San Francisco, CA) [online], 5 (8), pp.e3736. DOI: 10.7717/peerj.3736 .
Strzelecki,. J (2001). The elk at Tsar’s Route, Biebrza National Park (Poland). Wikimedia Commons. https://commons.wikimedia.org/wiki/File:Alces_alces_07(js),Biebrza_National_Park(Poland).jpg (CC BY-SA 3.0)
Tanner, A.L., and Leroux, S.J., 2015. Effect of Roadside Vegetation Cutting on Moose Browsing. PloS One [online], 10 (8), pp.e0133155. DOI: 10.1371/journal.pone.0133155 .
Tischler, K.B., Severud, W.J., Peterson, R.O. and Bump, J.K., 2019. Aquatic Macrophytes are Seasonally Important Dietary Resources for Moose. Diversity (Basel) [online], 11 (11), pp.209. DOI: 10.3390/d11110209 .
Touboul, J., Li, J., Patterson, D. and Levin, S., 2025. New challenges in spatial ecology: New challenges in spatial ecology. Japan Journal of Industrial and Applied Mathematics [online], 42 (4), pp.1695–1714. DOI: 10.1007/s13160-025-00720-7 .
Van Moorter, B., Rolandsen, C.M., Basille, M., Gaillard, J. and Börger, L., 2016. Movement is the glue connecting home ranges and habitat selection. The Journal of Animal Ecology [online], 85 (1), pp.21–31. DOI: 10.1111/1365-2656.12394 .
Wasser, S.K., Keim, J.L., Taper, M.L. and Lele, S.R., 2011. The influences of wolf predation, habitat loss, and human activity on caribou and moose in the Alberta oil sands. Frontiers in Ecology and the Environment [online], 9 (10), pp.546–551. DOI: 10.1890/100071 .
Appendix
Table A1. Summary statistics for each individual moose over the full three year study period including total number of tracking days and number of locations per moose.
Figure A2. Step length distributions for the two inferred behavioural states, restricted movement (blue) and travelling (red). Step length (x axis) compared against density (y axis).