Translator Disclaimer
1 April 2021 Don't go chasing the ghosts of the past: habitat selection and site fidelity during calving in an Arctic ungulate
Author Affiliations +
Abstract

Predator avoidance and food availability are both factors known to influence habitat selection and site fidelity around calving in caribou and reindeer. Here, we assess habitat selection and site fidelity during the calving period in the solitary, Arctic Svalbard reindeer Rangifer tarandus platyrhynchus, which is subject to limited predation risk and human disturbance. In this largely predator-free environment, we explore and discuss if habitat selection during the first week after calving is driven by food availability or remnants of anti-predatory behaviour. Based on GPS-collar data and ultrasound scanning (2009–2017; n = 134 individual-years) from two study areas, we estimated individual calving dates using recursive partitioning and first passage time and compared habitat selection and site fidelity of reproductive versus non-reproductive females. The K-select analysis suggested similar habitat selection during calving in reproductive and non-reproductive females. Female reindeer generally selected for lowland, flat habitats with high proportion of heath and moss tundra, i.e. habitats typically rich in terms of forage plants. Individuals producing a calf had significantly higher site fidelity in the calving period compared to the null model, and the mean distance between consecutive years' calving areas ranged between 1.5 and 3.9 km. Our study provides support for the prediction that in the absence of significant predation, ungulate calving site selection in the Arctic is mainly driven by the availability of spatially and temporally varying food resources.

Environmental conditions during gestation and after birth (Wolcott et al. 2015), including those affected by parental behavioural decisions, can influence reproductive success (McNamara and Houston 1986, Byers and Hogg 1995). For instance, choice of rearing habitat can be pivotal for offspring early survival (Refsnider and Janzen 2010). Animal habitat selection involves both responses in space and time to perceived risks and rewards and occur at multiple spatial and temporal scales (Mayor et al. 2009). For many female ungulates, the distribution of predators and the availability of forage resources are the two main factors explaining habitat selection in the calving period (Gustine et al. 2006). The ‘predation risk hypothesis’ states that females with dependent offspring trade good foraging locations for predator safe areas when the offspring is at its most sensitive stage to predation (Bowyer 1984, Brown et al. 1999), as seen in e.g. moose Alces alces (Tremblay et al. 2007), European red deer Cervus elaphus (Bonenfant et al. 2004) and reindeer and caribou Rangifer tarandus sp. (Mumma et al. 2017, Viejou et al. 2018). However, for many ungulates, predation risk alone cannot explain female habitat selection around the period of birth (Ruckstuhl and Neuhaus 2002). As the offspring grows, nutritional requirements increase (Cook et al. 2004), and access to foraging areas with high quality and quantity of forage is essential (Ruckstuhl and Neuhaus 2002). In predator-free environments, one could therefore anticipate that forage resources are the main determinant of habitat selection, and that the habitat selection patterns are independent of reproductive status. However, even in the absence of predators, relict anti-predator behaviour, has been observed to affect habitat selection in ungulates (Byers 1998, Mahoney and Schaefer 2002).

Site fidelity, i.e. the tendency of an individual to return to a previously used area, is often strong in ungulates (Gunn and Miller 1986, Tremblay et al. 2007). The past experience of spatial predation patterns, food resource distribution and positive reproductive outcomes may encourage females to return to the same calving area (i.e. philopatric behaviour) and thus help them maximize reproductive success (Switzer 1993, Mettke-Hofmann 2017). However, it is not clear whether calving site fidelity is mainly driven by predictability of reproductive success (i.e. the expectation that future outcomes will be the same as the current outcome, if you do the same) or by the assessment of environmental cues, i.e. temporally or spatially varying costs and benefits. In general, site fidelity may be weaker the less predictable resources are in space and time (Arthur et al. 2015). For instance, terrestrial Arctic ecosystems are characterized by long, harsh winters, short summers and large seasonal variability in forage availability. During spring and early summer, the landscape characteristics typically change from total snow cover to a mosaic of snow covered and exposed vegetation, and eventually, a ‘greening’ landscape. Because of large annual variation in timing of snowmelt and spring onset (Pedersen et al. 2016, Schmidt et al. 2019), the spatial distribution and availability of forage is likely to be highly variable both within and between years. Low spatiotemporal predictability of food availability in potential calving areas may therefore lead to reduced calving site fidelity in the Arctic compared to more predictable environments (Callaghan et al. 2011). Thus, particularly low calving site fidelity could be anticipated in Arctic ungulates experiencing low predation pressure, such as the wild Svalbard reindeer Rangifer tarandus platyrhynchus (Tyler 1987, Derocher et al. 2000). The medium sized predator, the Arctic fox Vulpes lagopus is a common scavenger of Svalbard reindeer (Prestrud 1992, Eide et al. 2005) and capable of killing new-born calves (Prestrud 1992), but there is very limited evidence of such predation events or attempts (Tyler 1986, 1987, Prestrud 1992). Polar bears Ursus maritimus occur on the Svalbard tundra during parts of the year but have very rarely been observed killing adult reindeer (and never calves; Derocher et al. 2000).

In this observational study, we used a GPS-collar dataset with 134 individual-years from Svalbard reindeer females (2009–2017) to estimate individual calving dates and analyse seasonal habitat selection and site fidelity patterns of reproductive versus non-reproductive individuals. We explored if, in an environment with negligible predation risk, habitat selection of female reindeer during the calving period is driven by selection for areas with high food availability (Hamel and Côté 2008) or remnants of anti-predatory behaviour during this most sensitive life stage for the Svalbard reindeer.

Material and methods

Study area

This study was conducted on Spitsbergen in the Arctic Svalbard archipelago, and included two study areas that differ in terrain and climate characteristics – Nordenskiöld Land (78°N, 15–16°E; middle Arctic tundra zone; Elvebakk 1999) and West-Spitsbergen (78°N, 11–12°E; High-Arctic tundra zone) (Fig. 1). The Nordenskiöld Land study area (hereafter ‘inland area'; approx. 150 km2) consists of wide u-shaped valleys connected by smaller valleys through high elevation passes. The area contains a wide variety of habitat types, ranging from continuous vegetation of different types in the valleys to barren ground at higher elevation. Marshes, wetlands and moist moss tundra dominate the lowland valley bottoms, while various types of moss tundra and heaths cover the foothills and slopes (Johansen et al. 2012). The West-Spitsbergen study area (hereafter ‘coastal area'; approx. 296 km2) consists of three peninsulas, which are separated by open sea and large tidewater glaciers. These peninsulas are characterized by coastal plains with an abrupt shift to steep and rugged terrain with alpine mountains. The vegetation cover in the coastal plains is discontinuous and consists of heath, open ridge communities and small pockets of different types of moss tundra (Johansen et al. 2012).

The study areas differ in terms of climate and weather variability, including length of the period with snow cover and the amount of precipitation. The coastal area (Ny-Ålesund weather station) has on average six weeks longer period of snow cover than the inland area (Svalbard airport weather station) (2009–2017; Norwegian Meterological Inst. 2018), and the timing of onset of the growing season is also overall later in Ny-Ålesund (Karlsen et al. 2014). Furthermore, the mean annual precipitation is more than double in Ny-Ålesund (541.9 ± 122.8 mm) than at Svalbard airport (209.7 ± 41.6 mm) (2009–2017; downloaded from < www.seklima.met.no>).

Study species

Svalbard reindeer are non-migratory and appear solitary or in small, often sexually segregated groups (Loe et al. 2006). Individuals occupy seasonal home ranges that are small compared to the migratory reindeer and caribou herds elsewhere in the Arctic (Tyler and øritsland 1989). However, partial seasonal migration can occur, for instance in severe winters (i.e. with multiple icing events) and when forage resources are scarce (Hansen et al. 2010, Loe et al. 2016). The population fluctuations are mainly driven by winter weather variability and density-dependence, which operates through competition for food and gastrointestinal parasitism (Albon et al. 2002, Kohler and Aanes 2004, Tyler et al. 2008, Hansen et al. 2013).

Reindeer GPS-data

Reindeer capture

Female Svalbard reindeer were marked with GPS-collars to obtain data on their habitat and space use. The reindeer were captured during winter (February to April 2009–2017) using a handheld net from a pair of snowmobiles (Omsjoe et al. 2009). In total, 84 and 51 individual female reindeer from the inland and coastal areas, respectively, were fitted with GPS-collars (Vectronic ‘store-on-board’ collars in the inland study area (2009–2015), Followit satellite collars in the inland study area (2016–2017) and the coastal study area (2014–2017)) during the study period. All animals were handled according to protocols approved by the Governor of Svalbard and the Norwegian Food Safety Authority.

Figure 1.

Locations of GPS-marked Svalbard reindeer females the seven days after calving in summers 2009–2017 in the inland study areas of Nordenskiöld Land (left panel) and coastal study areas of West-Spitsbergen (right panel). Light-grey areas are merged spring home ranges (95% KernelUD, 15 May–30 June) for all females with calves in the period 2009–2017. Blue stars = estimated individual calving sites from GPS movement metrics (RP and FPT). Circles = all GPS-positions seven days after calving for females with calves (red), and seven days after median calving day for females without calves (orange).

img-z3-1_01.jpg

Reproductive status

We recorded reproductive status of the GPS-collared females using ultrasound and/or progesterone blood sampling during capture and as the presence or absence of a calf at heel during late summer censuses in July and August (Albon et al. 2017 and Moullec et al. 2017 for description of protocols). Thus, reproductive females were defined as pregnant in spring and observed with a calf in the late summer census, while non-reproductive females were defined as not pregnant in spring and not observed with a calf in the summer census.

Pre-processing of GPS-data

To remove errors and outliers from the GPS-dataset, we followed the screening protocols by Bjørneraas et al. (2012), which reduced the number of unrealistic distances between GPS-positions (median > 100 km, mean > 10 km) and turning angles (> 1.5 km h–1). The conservative movement thresholds were chosen because they handle the tradeoff between removing location errors and retaining sufficient GPS-positions in the dataset. In addition, all GPS-positions with a horizontal dilution of precision (HDOP) more than 10 m, were excluded to have similar spatial precision of GPS-positions (Recio et al. 2011). In total, these screening procedures removed 7% of the GPS-positions. The Vectronic ‘store-on-board’ collars recorded positions every 1–2 h, while the Followit satellite collars recorded positions once every eight hour. For consistency, all datasets were standardized to one position every eight hour (nearest GPS-position to 08:00, 16:00 and 00:00, Greenwich Mean Time). To reduce errors when estimating calving sites for females, we excluded individual-years if more than 60% of GPS-positions were missing in the pre- and post-calving period (15 May–30 June). On average, the remaining individual-year trajectories had a mean successful fix rate of 97% (93% for Follow-it satellite link service, 98% for Vectronic store-on-board). The resulting dataset contained data from 134 individual-years (Ninland = 98, Ncoastal = 36) available for analysis (Table 1).

Environmental data

To study female habitat selection during the calving period, we selected 11 environmental variables based on literature about habitat selection in calving ranges elsewhere, the ecology of Svalbard reindeer and availability of digital spatial layers for the study areas (Table 2, for correlations among variables see Supporting information). We resampled all raster layers to the same spatial extent and resolution (30 × 30 m) using ArcMap 10.6.1 (ESRI 2011) and R Studio ver. 1.0.143 (< www.r-project.org>). For any raster layer with missing values within an individual's home range, missing values were replaced with the mean of existing values for the given home range. This way missing values were given less weight in the habitat selection analyses.

Table 1.

Sample size overview of GPS-collared Svalbard reindeer females for each study year and area (total Ninland=98; Ncoastal=36; Fig. 1 for study areas) used in the habitat selection analyses. Calf status reflects the number of females with (1) and without (0) a calf at heel in late summer.

img-z4-2_01.gif

Vegetation variables

We extracted vegetation variables from the digital vegetation map by Johansen et al. (2012). Because several of the initial vegetation classes were ambiguous, based on the overall validation statistics, we regrouped the vegetation classes to three coarse habitat types reflecting available foraging habitats (termed ‘moss tundra’, ‘heath’ and ‘barren'). Moss tundra (e.g. meadows, various types of wetlands and bird-cliff vegetation) reflects the most biomass-rich foraging habitats (mean biomass 330 g m–2), while heath reflects less productive foraging habitats that constitute open heath communities and grasslands (mean biomass 225 g m–2). Barren consist of areas with non-vegetated to sparsely vegetated areas, gravel and polar desert (mean biomass 28 g m–2) (details in the Supporting information). All three habitat types are present in the inland and coastal areas in varying extents (Supporting information). When calculating the vegetation variables, we applied a moving average so that each 30 × 30 m pixel had the mean value based on the pixels in its immediate surroundings (3 × 3 pixels; 90 × 90 m).

Onset of the growing season

Annual raster maps describing the timing of the ‘onset of the growing season’ were available for the period 2009–2016 (Karlsen et al. 2014 for details). The onset of the growing season was defined as the day when the NDVI-value of a pixel first passed 70% of the same pixel's annual maximum NDVI-value (termed spring onset). This proxy for onset of growing season in Svalbard correlates well with the flowering of polar willow Salix polaris in Nordenskiöld Land, which is an important foraging plant for Svalbard reindeer (Bjørkvoll et al. 2009). For the habitat selection analyses, we calculated the mean onset of the growing season between 2009 and 2016 as we predicted reindeer to select calving areas with generally earlier growing season onset.

Terrain variables

Altitude, slope in radians and aspect in degrees were extracted from a Digital elevation model (DEM, 5–10 m estimated uncertainty, Norwegian Polar Inst. 2014). From the DEM, we also calculated three indices relevant to reindeer habitat selection: 1) terrain wetness, which represents an inverse measure of available dry habitat, was calculated following the ‘the topographical wetness index' by Beven and Kirkby (1979). 2) The amount of incoming solar radiation, which affects snowmelt and plants' growing conditions (Pedersen et al. 2017), was calculated following the ‘the heatload index' by Parker (1988). 3) Terrain ruggedness, which is a measure of topographical heterogeneity, was calculated following the ‘vector ruggedness measure’ by Sappington et al. (2007).

Predation risk variable

Although predation by Arctic fox on reindeer calves is considered rare, observations of foxes harassing reindeer mothers and calves exist (Tyler 1986, 1987, Prestrud 1992). To investigate to what extent spatial variation in fox breeding den presence may influence Svalbard reindeer habitat selection in the calving period, we calculated a raster layer displaying the linear distance from each pixel to a known fox den in Svalbard (n = 81) and used this variable (termed foxdens) as a proxy for predation risk for reindeer calves. The position of fox breeding dens was obtained from the database ‘Arctic fox dens on Svalbard’ database that contain information on den sites from 1982 till today (Fuglei unpubl.). Arctic fox dens are used over generations and even for centuries (Prestrud 1992).

Statistical analyses

Estimation of calving day and site

To estimate calving day, and hence individual calving site from GPS-data, we used two methods based on movement metrics: 1) recursive partitioning, which uses movement rate and net displacement to capture distinct calving behaviours before, during and after birth in caribou (RP, Rudolph and Drapeau 2010), and 2) first passage time, which calculates time spent in a certain area, commonly applied in studies on foraging ecology of species (FPT, Fauchald and Tveraa 2003). For the RP method, we defined individual calving day as the day of the year where a recursive partitioning line (identifying a significant change from high to low movement rate) coincided with the minimum point in movement rate in the pre- and post-calving period (here defined as 15 May–30 June). This was followed by constant net displacement indicating that the female remained stationary. For the FPT method, we defined calving day as the date with the highest FPT value after 30 May, which corresponds to the date when females with calves required the longest time to move through a circle with radius 100 m, i.e. the date with most sedentary behaviour indicating calving. The RP and FPT analyses were computed using the Zoo and AdehabitatLT packages in R (Calenge 2016; see the Supporting information for an individual-year example). For both movement metrics, data for each individual-year were categorized according to how precisely calving day was determined. The categories were: 1) one-day certainty (n=50), 2) 1–3 days certainty (calving day recorded as the first day of minimum movement rate after the recursive partitioning line, n=16) and 3) no clear calving day (n = 14). Only data from females in category 1 was used in the subsequent habitat selection analyses. We found that the calving dates identified in the RP method correlated well with the calving dates identified from the FPT analyses in most cases (70%), but when these dates differed, recursive partitioning was used since this is a published calving day estimation method (Rudolph and Drapeau 2010).

Table 2.

Overview of environmental variables used to assess Svalbard reindeer calving habitat selection, with observed (reindeer/caribou in general) and expected effects on calving habitat selection given two main constraints (forage availability and predator avoidance). Prior to statistical analysis, all raster layers were re-sampled from their original resolution to a spatial resolution of 30 × 30 m to match the spatial resolution of the digital vegetation map by Johansen et al. (2012).

img-z5-2_01.gif

Defining used and available habitats

We applied third-order habitat selection to investigate how habitat components within the home range are utilized in the calving period (Johnson 1980, Manly et al. 2002). Used and available habitat were defined for each individual-year separately to best match the calving period for each female (design III, Manly et al. 2002). For females with both a calf at heel in late summer censuses and an estimated calving day (n = 48), we defined the calving area as the location of all GPS-positions the first seven days after calving (Supporting information). We chose this period because the results of K-selection analyses on shorter time intervals after calving (location of females one, two, three and five-days post-parturition) appeared qualitatively similar. To compare habitat selection between reproductive and non-reproductive females, the used area for non-reproductive females in spring field surveys (n = 30) was defined as the location of all GPS-positions the first seven days after median calving day in the same population and year. For every individual-year in the two reproductive groups, the available habitat was defined as the spring home range (calculated using 95% Kernel's utilization distribution) from 15 May to 30 June, which also included the calving period.

Habitat selection in the calving area

To assess individual Svalbard reindeer calving habitat selection, we applied multivariate K-select analysis where environmental variables define a multi-dimensional niche space (Calenge et al. 2005). The K-select analysis is an exploratory factorial analysis that applies marginality vectors, i.e. the difference between vectors of an individual's mean utilised habitat and its mean available habitat in centred and non-centred principal component analyses. The size of the marginality vector for an individual is proportionate to the strength of habitat selection. The eigenvalues indicate the amount of mean marginality explained by each factorial axis. The K-select analysis is advantageous because correlation of habitat variables does not affect the results, both categorical and quantitative variables can be used, and the analysis graphically displays individual differences within populations. For each individual-year, the mean used habitat in the calving period is compared to mean available habitat. We separated individual-years into inland (Nwith_calf = 36, Nno_ calf = 20) and coastal study areas (Nwith_calf = 12, Nno_calf = 10). To test whether habitat selected was non-random for these study populations and reproductive groups, we applied randomization tests on the marginality for each individual-year using a Bonferroni correction (n = 1000 repetitions, Calenge et al. 2005).

To summarize the importance of each of the 11 habitat variables across females of different reproductive status, we calculated the mean marginality vector length of each variable, i.e. the average across individuals of the difference between mean used and the mean available for the variable. The higher the absolute value of these mean marginality estimates, the further the reindeer departed on average from the mean available habitat of the given variable. A positive coefficient of a variable indicates that the females selected calving areas where the value of the variable is higher than the mean of the available habitat in spring home ranges, while negative coefficients indicate selection for areas with lower than the mean variable value. If the coefficient is close to zero, the mean variable value in the calving area is similar to the mean variable value in the spring home ranges (i.e. used is the same as available habitat). To test whether habitat selection differed significantly between reproductive and non-reproductive females, linear mixed effect models were performed on each environmental variable separately in R using the package nlme. We used the marginality coefficient of each habitat variable as the dependent variable, reproductive status as a predictor variable, while animal id were used as a random effect.

Site fidelity in the calving and rearing period

We analyzed site fidelity using the approach from Schaefer et al. (2000), where seasonal movement properties of the study animal define the scale of site fidelity. This allowed us to investigate how close Svalbard reindeer females is to a location that they occupied in previous years without setting an arbitrary spatial scale, an advantage when making comparisons between individuals. Distances between paired GPS-positions of a female obtained one year apart or more, but on the same day of year, were calculated for the whole year, and then aggregated into mean weekly distances for every individual. Data were available for 22 pairs of individual-years and the females had these possible reproductive outcomes in year t and year t >– 1: 1) calf–calf (n = 10), 2) no calf–no calf (n = 12), 3) no calf in one year and a calf in the other year (n = 8). To assess how site fidelity in the calving period differed from other seasons, the mean and 95% confidence intervals of individual distances were calculated for each week of the year for each reproductive group. We divided the spring and summer seasons into periods to ease group comparisons: calving (2–19 June) and rearing (early summer 20–30 June; midsummer 1–31 July; late summer 1–31 August; following temporal scales defined by Loe et al. (2006). We combined inland and coastal study populations due to the low sample size from the coastal populations. To investigate differences in site fidelity during the calving and rearing period between reproductive groups, we used the non-parametric Wilcoxon rank-sum test to account for non-normality of the data.

To assess site fidelity in different seasons we developed a null model. First, we defined an individual's range as the space denoted by all GPS-positions in year t and year t > 1. Only pairs of individual-years with GPS-positions covering > 300 days per year were included to ensure all seasons were covered. Within this individual's annual range, we randomly sampled 1000 pairs of GPS-positions and calculated the distance between positions in each pair. These 1000 randomized distances were then averaged for pairs of individual-years to create the null model (representing the distribution of expected distance between any two GPS-positions). Significant site fidelity occurred in periods when the distance between the average of ‘real pairs' of individual-years was lower than the lower 95% confidence limit of the average ‘randomized pairs’ of individual-years in the null model.

All statistical analyses were performed in R ver. 1.456 < www.r-project.org>.

Results

Estimation of calving day and site

Estimated calving dates for GPS-collared female Svalbard reindeer ranged from 5 to 17 June with median date of 8 June (Inland: median=7 June, SD=4 days, Coastal: median=11 June, SD=7 days). In the inland study area, most calving sites were aggregated in the innermost upland areas of Colesdalen (40%, n=19) and Semmeldalen (34%, n=16) (Fig. 1). In the coastal study area, no apparent pattern existed, all calving sites were distributed on the coastal flats and in the foothills (n=18).

Figure 2.

Results of the K-select analysis for Svalbard reindeer female habitat selection during the calving period in (a) inland and (b) coastal populations (Ncoastal = 22, Ninland = 56). For both figures: 1) variable loadings on the first two factorial axes related to vegetation, terrain and calf predation predictor variables, 2) eigenvalues in % for 11 factorial axes (> 70% contribution shown in black), 3) uncentered (middle panel) and 4) centred (right panel) projections of the marginality vectors on the first and second factorial planes for individual females with calves (red arrows) and females without calves (orange arrows). In the diagrams, the tip of the arrow indicates used habitat (7 days post-calving for females with calves and 7 days after median calving day for females without calves), while the base of arrow indicates available habitat (individual spring home ranges, 15 May–30 June). The length of the arrow indicates marginality, i.e. the strength of habitat selection. The variable ‘onset of the growing season' is denoted as ‘springonset' on the figures. and the variable ‘distance to fox dens' is denoted as ‘foxdens’ in the figures

img-z7-10_01.jpg

Habitat selection in the calving areas

In the inland area, the first axis of the K-select analysis explained 44.4% of the variation in average marginality, while the second and third axis explained 17% and 15% of the variation, respectively (Fig. 2a). The first factorial axis was positively related to variables associated with forage such as the proportion of moss tundra and terrain wetness, and negatively related to ruggedness, slope steepness, altitude and proportion of heath and barren areas. The second factorial axis was positively related to the onset of the growing season, heatload, aspect and distance to fox dens (Fig. 2a, Supporting information). Thus, as indicated by the direction of the arrows, most female reindeer in the inland populations selected areas in flat, lowland, moss tundra landscapes. Disproportionate use of habitat during the calving period, compared to available habitat, occurred for 14 females in the reproductive group and 7 females in the non-reproductive group, as indicated by the results of the randomization tests (Supporting information). Non-reproductive females selected significantly more frequently for heath in their used habitat during the calving period compared to reproductive females (F1,54 = 4.30, p = 0.043; Fig. 3a), while all other environmental variables were selected similarly for by both reproductive groups. In combination, this implies that the majority of females with calves select for calving habitat similar to their available area in the first week after birth, and that non-reproductive females are more frequently found on heath compared to reproductive females in the calving period.

In the coastal area, the first axis of the K-select analysis explained 39.9% of the variation in the total marginality, while the second and third axis explained 23.6% and 13.2% of the variation respectively (Fig. 2b). The first factorial axis was positively related to proportion of heath and terrain wetness, and negatively related to altitude, terrain ruggedness, slope steepness and barren areas. The second factorial axis indicated selection towards areas with high proportion of moss tundra and high wetness levels (Fig. 2b, Supporting information). Thus, most female reindeer in the coastal populations selected areas in flat, lowland, heath tundra landscapes. Habitat selection was significantly non-random for nine (four females with calves and five females without calves) out of 22 females during the calving period at both the 10% and 5% level (Supporting information). As for the inland females, the importance of environmental variables was similar for reproductive and non-reproductive females, but higher individual variability in habitat use during the calving period existed. Thus, there was a tendency for more consistent selection for habitats with high plant biomass production as indicated by high proportion of moss tundra in the inland than the coastal area.

Figure 3.

Mean marginality vector lengths from the K-select analysis of habitat selection in Svalbard reindeer females 7 days post-calving for females with calves (red boxes) and 7 days after median calving day for females without calves (orange boxes) in (a) inland and (b) coastal populations. The variable ‘onset of the growing season' is denoted as ‘springonset’ on the figures. For the coastal populations, the variable springonset is marked with an asterisk because mixed effect models indicated a significant difference between reproductive and non-reproductive females.

img-z8-6_01.jpg

Site fidelity in the calving and rearing period

Svalbard reindeer females displayed varying degrees of site fidelity to calving areas, contingent on time of the year and their reproductive status (Table 3, Fig. 4). After the onset of calving (2–19 June), females with a calf in two consecutive years displayed a significantly shorter distance between GPS-positions compared to both the null model than to females with calf in one year but not the other. This provides indicative evidence for site fidelity to calving areas. The mean distances between consecutive calving areas for females with calves in two consecutive years were 1.5–3.9 km, suggesting that calving site fidelity operated on a relatively coarse spatial scale relative to the annual home range size of 50–65 km2 (resembling a diameter of 8–9 km if circular, n = 134). Site fidelity remained high throughout the entire summer in females with calves in two consecutive years compared to the null model (Fig. 4). Females with no calf in either of the years, and females with no calves in one of the years displayed significant site fidelity, but later in the summer.

Table 3.

Mean distances (km) between paired GPS-positions of females, obtained one year apart on the same day of the year, during the summer season across Svalbard reindeer females of different reproductive status. Distances that are significantly shorter than the null model (5.3 km) are marked with *.

img-z9-2_01.gif

Discussion

In this study we demonstrated that habitat selection of female Svalbard reindeer during the calving period is largely independent of reproductive status. As expected in this virtually predator-free environment, individual reindeer generally selected for flat, lowland habitats associated with high forage availability. There was no evidence of selection for habitats assumed to reduce predation risk for calves (i.e. proximity to fox dens, terrain ruggedness or steep slopes). This indicates that low predation pressure allows females to choose their calving location based only on forage availability. Nevertheless, we found indications of site fidelity to the calving area, but at a coarse scale relative to the size of individual home ranges. The apparent paradox of a significant calving site fidelity despite similar habitat selection as non-reproductive females may be explained by remnants of anti-predatory behaviour or memory of rich summer foraging resources and past reproductive success (Byers 1998, Mahoney and Schaefer 2002).

Figure 4.

Left panel (A–C): the black line and coloured areas represent the mean weekly distances and 95% confidence intervals between GPS-positions obtained one year apart but in the same week for individual reindeer in the same reproductive group (red = females with calf at heel in both years, orange = females without calf in both years, grey = females with calf at heel in only one of the two consecutive years). The grey horizontal line is the null model (i.e. randomized paired GPS-positions for each female reindeer in year t and t + 1). The time-periods separated with stippled vertical lines are: P-C = pre-calving (15 May–2 June), C = calving (2–19 June), E-S = early summer (20–30 June), M-S = mid-summer (1–31 July), L-S = late summer (1–31 August) (Loe et al. 2006). Right panel (D–G): boxplot showing the group medians and null model for each reproductive group in the calving and rearing periods. Asterisk indicates a significant difference in mean between reproductive groups from Wilcoxon rank-sum test.

img-z9-9_01.jpg

Habitat selection in calving areas

Adult females selected moss and heath tundra during the calving period in the coastal and inland area, respectively (Fig. 2, 3). These habitat classes contain vegetation types characterized by high amounts of foraging plant biomass (Johansen and Tømmervik 2014; Supporting information) and food quality during the calving period (Bjune 2000, Peignier et al. 2019). Similar habitat selection is also observed in other High-Arctic caribou populations such as the Alaskan barren-ground caribou Rangifer t. granti (Kelleyhouse 2001). Common for the barren-ground caribou' calving areas are that they contain high green biomass either during calving or peak lactation, and often provide better forage than elsewhere in their seasonal ranges (Russell et al. 2002). The selection for moss tundra and heath habitats in Svalbard reindeer likely reflect their nutritional requirements and the need to be in proximity to areas with good forage quality and quantity as the snow melts. In our study, heath was the only variable where habitat use significantly differed between reproductive and non-reproductive females (Fig. 2, 3). Energy cost during the calving period for reproductive females is higher than for non-reproductive females due to the additional need for lactation and calf nursing, which makes additional movement costly. Ridges with heath become earlier snow-free, but also contains less biomass of relevant foraging plants compared to foraging plants in moss tundra. This suggests that non-reproductive females can be more opportunistic in their foraging and movement behaviour, and thus use a larger foraging area during early spring compared to females with calves.

The slight differences in female habitat selection between our inland and coastal study areas (Fig. 2, 3), may reflect differences in available vegetation, snow melt patterns and onset of the growing season. Heath also becomes earlier snow-free on the coastal plains, which may explain why available pockets of nutrient-rich moss tundra near bird cliffs were not selected as calving areas. The inland study area becomes earlier snow-free, and patches of moss tundra can be exposed as early as the calving period. In addition, the onset of the growing season is more consistent between years in the foothills of the inland calving areas compared to the coastal plains (Karlsen et al. 2014). Therefore, the selection for heath in coastal areas and moss tundra in inland study area may simply reflect habitat availability during the snow melting period.

As assumed from the low predation pressure in Svalbard (Tyler 1986, 1987, Prestrud 1992), there was no evidence for selection of habitat variables associated with predation risks for reindeer calves (e.g. high elevations, steep slopes, rugged terrain and large distance from fox dens; Fig. 2, 3) at our study scale (i.e. foraging area). This contrast other reindeer and caribou populations that seek habitats with low predator-densities and high forage quality and quantity during the calving period to maximize calf survival (Nobert et al. 2016, Viejou et al. 2018). Our results correspond, however, to habitat selection patterns of other ungulates, such as the roe deer in the study of (Dupke et al. 2017), where food availability rather than lynx predation risk governs habitat selection on multiple scales. However, however, the Arctic is also experiencing altered predator–prey relations under climate change (Post et al. 2009), and both Arctic fox (Prestrud 1992) and polar bears (Derocher et al. 2000) are capable of killing reindeer (calves and adults, repectively), thus the current knowledge about predator–prey interactions in Svalbard (Tyler 1986, 1987, Prestrud 1992, Derocher et al. 2000) may change. The expected expansion of polar bear range use onto land (Rode et al. 2015) might lead to increased use of alternative prey resources, such as reindeer (Stempniewicz et al. 2013, Kavan 2018).

Site fidelity in the calving and rearing period

We found that parturient females had significantly higher site fidelity during the calving period (2–19 June) than females without calves in the same period (Fig. 4), which is supported by previous studies (Tyler 1986, Hansen et al. 2010). The distances between individual calving areas were of similar order of magnitude as for predator-free moose (Tremblay et al. 2007), but on average closer than for e.g. North-American migratory caribou R. t. caribou (8.7 km (Nobert et al. 2016); 3.6 km (Popp et al. 2008)) and sedentary caribou (6.7 km; Schaefer et al. 2000) in the post-calving period.

There was no significant difference in habitat selection between reproductive and non-reproductive females. Still, many of the reproductive females returned to the same calving areas the next year. This fidelity behavior, with no apparent benefit, is a paradox and could likely be explained by innate conservative traits, as seen in other ungulate populations without predators (Byers 1998) and populations with low incidental predation rates (Qin 2011, LaFontaine et al. 2017), or alternatively be associated with aspects of resources at a more detailed spatial level than what we could detect in our study. Remnants of anti-predatory behavior in predator-free or nearly predator-free environments can persist for thousands of years (Byers 1998), especially if these innate conservative traits are not too costly to maintain (Neill 1990). For instance, caribou in Newfoundland continues to migrate in the absence of wolves (Mahoney and Schaefer 2002), and bed-site selection of roe deer with fawns was still influenced by anti-predatory strategies for decades after eradication of predators (Qin 2011).

Previous reproductive success could also play a role in calving site selection in subsequent years for Svalbard reindeer, and may explain the variation in site fidelity between individuals in different reproductive groups. Reproductive success is influenced by food availability in the calving site, and the slight difference in site fidelity between reproductive groups during calving suggests that relying on prior knowledge of foraging resources may be especially important for females with calves. Remembering where the best foraging areas are located and being adjacent to these can therefore be energetically advantageous (Gunn and Miller 1986). The increasing site fidelity pattern over the summer aligns with studies from North American caribou populations (Schaefer et al. 2000, Popp et al. 2008) and the fact that site fidelity is linked to predictability in resource availability (Switzer 1993), which increases after snowmelt and over the growing season. The foraging landscape during the summer rearing period is more predictable than during calving, and it is possible that females are ‘homing in' on an area that are important in foraging resources once the snow disappears. Although memory of rich summer foraging patches and previous reproductive success may be a plausible contender to the ‘ghost of predators past’ hypothesis we cannot rule out one over the other. More research needs to be done to effectively evaluate these hypotheses, for instance by investigating the role of reproductive success in selection and fidelity of calving sites.

Conclusion

We demonstrate that female reindeer in this High-Arctic environment select habitats associated with good forage conditions over predation-risk during the critical stage after calving. It is possible that the reindeer rely on former knowledge of the best food patches and place themselves accordingly before the onset of the growing season. The apparent paradox of a significant calving site fidelity, despite similar habitat selection among females with different reproductive status, may be explained by innate conservative traits (Byers 1998) or relict anti-predatory behaviour (Mahoney and Schaefer 2002). This explorative study provides a baseline for the continuation of hypothesis-based studies on calving habitat selection where individual fitness components are linked to habitat selection and site fidelity at the critical stage of calving.

Acknowledgements

We thank Rudolph Tyler, Ivar Herfindal and Arnaud Tarroux for contributing to the R scripts, Stein Rune Karlsen for NDVI maps made available through MOSJ, Hans Tømmervik for providing biomass values, Isabell Eischeid for processing the digital raster layers Mads Forchhammer for Master thesis supervision and Dorothee Ehrich and Karen Lone for constructive comments to an earlier version of the manuscript. This study is a contribution from Climate-Ecological Observatory for Arctic Tundra (COAT, < www.coat.no>). Norwegian Food Safety Authority gave permits for conducting the studies (FOTS ID 6047, 14980, 19061, 22808).

Funding – The Norwegian Polar Institute, Norwegian Institute for Nature Research, Norwegian University for Life Sciences, Norwegian University of Science and Technology, The James Hutton Institute, UiT – Arctic University of Norway, Tromsø Research Foundation, Svalbard Environmental Protection Fund and Research Council Norway (FRIMEDBIO 276080, SFF 223257, KLIMAFORSK 267613) provided funding for the study.

References

1.

Albon, S. D. et al. 2002. The role of parasites in the dynamics of a reindeer population. – Proc. R. Soc. B 269: 1625–1632. Google Scholar

2.

Albon, S. D. et al. 2017. Contrasting effects of summer and winter warming on body mass explain population dynamics in a food-limited Arctic herbivore. – Global Change Biol. 23: 1374–1389. Google Scholar

3.

Arthur, B. et al. 2015. Return customers: foraging site fidelity and the effect of environmental variability in wide-ranging antarctic fur seals. – PLoS One 10: e0179322. Google Scholar

4.

Beven, K. J. and Kirkby, M. J. 1979. A physically based, variable contributing area model of basin hydrology/Un modèle à base physique de zone d'appel variable de l'hydrologie du bassin versant. – Hydrol. Sci. Bull. 24: 43–69. Google Scholar

5.

Bjørkvoll, E. et al. 2009. Seasonal and interannual dietary variation during winter in female Svalbard reindeer Rangifer tarandus platyrhynchus. – Arct. Antarct. Alp. Res. 41: 88–96. Google Scholar

6.

Bjorneraas, K. et al. 2012. Habitat quality influences population distribution, individual space use and functional responses in habitat selection by a large herbivore. – Oecologia 168: 231–243. Google Scholar

7.

Bjune, A. E. 2000. Pollen analysis of faeces as a method of demonstrating seasonal variations in the diet of Svalbard reindeer Rangifer tarandus platyrhynchus. – Polar Res. 19: 183–192. Google Scholar

8.

Bonenfant, C. et al. 2004. Multiple causes of sexual segregation in European red deer: enlightenments from varying breeding phenology at high and low latitude. – Proc. R. Soc. B 271: 883–892. Google Scholar

9.

Bowyer, R. T. 1984. Sexual segregation in southern mule deer. – J. Mammal. 65: 410–417. Google Scholar

10.

Brown, J. S. et al. 1999. The ecology of fear: optimal foraging, game theory and trophic interactions. – J. Mammal. 80: 385–399. Google Scholar

11.

Byers, J. A. 1998. American pronghorn: social adaptations and the ghosts of predators past. – Univ. of Chicago Press. Google Scholar

12.

Byers, J. A. and Hogg, J. T. 1995. Environmental effects on prenatal growth rate in pronghorn and bighorn: further evidence for energy constraint on sex-biased maternal expenditure. – Behav. Ecol. 6: 451–457. Google Scholar

13.

Calenge, C. 2016. Analysis of animal movements in R: the adehabitatLT Package. – Office National de la Chasse et de la Faune Sauvage Saint Benoist, pp. 1–93. Google Scholar

14.

Calenge, C. et al. 2005. K-select analysis: a new method to analyse habitat selection in radio-tracking studies. – Ecol. Model. 186: 143–153. Google Scholar

15.

Callaghan, T. V. et al. 2011. Multiple effects of changes in arctic snow cover. – Ambio 40: 32–45. Google Scholar

16.

Cook, J. G. et al. 2004. Effects of summer–autumn nutrition and parturition date on reproduction and survival of elk. – Wildl. Monogr. 155: 1–61. Google Scholar

17.

DeCesare, N. J. et al. 2014. Linking habitat selection and predation risk to spatial variation in survival. – J. Anim. Ecol. 83: 343–352. Google Scholar

18.

Derocher, A. E. et al. 2000. Predation of Svalbard reindeer by polar bears. – Polar Biol. 23: 675–678. Google Scholar

19.

Dupke, C. et al. 2017. Habitat selection by a large herbivore at multiple spatial and temporal scales is primarily governed by food resources. – Ecography 40: 1014–1027. Google Scholar

20.

Eide, N. et al. 2005. Dietary responses of Arctic foxes Alopex lagopus to changing prey availability across an Arctic landscape. – Wildl. Biol. 112: 109–121. Google Scholar

21.

Elvebakk, A. 1999. Bioclimatic delimitation and subdivision of the Arctic. – In: Nordal, I. and Razzhivin, V. Y. (eds), The species concept in the high north – a panarctic flora initiative. Norske Videnskaps-Akademi, pp. 81–112. Google Scholar

22.

ESRI 2011. ArcGIS Desktop: Release 10. – Environmental Systems Res. Inst. Google Scholar

23.

Fauchald, P. and Tveraa, T. 2003. Using first-passage time in the analysis of area-restricted search and habitat selection. – Ecology 84: 282–288. Google Scholar

24.

Gunn, A. and Miller, F. 1986. Traditional behaviour and fidelity to caribou calving grounds by barren-ground caribou. – Rangifer 6: 151–158. Google Scholar

25.

Gustine, D. D. et al. 2006. Calf survival of woodland caribou in a multi-predator ecosystem. – Wildl. Monogr. 165: 1–32. Google Scholar

26.

Hamel, S. and Côté, S. D. 2008. Trade-offs in activity budget in an alpine ungulate: contrasting lactating and nonlactating females. – Anim. Behav. 75: 217–227. Google Scholar

27.

Hansen, B. B. et al. 2010. Partial seasonal migration in high-arctic Svalbard reindeer Rangifer tarandus platyrhynchus. – Can. J. Zool. 88: 1202–1209. Google Scholar

28.

Hansen, B. B. et al. 2009. Winter habitat–space use in a large arctic herbivore facing contrasting forage abundance. – Polar Biol 32: 971–984. Google Scholar

29.

Hansen, B. B. et al. 2013. Climate events synchronize the dynamics of a resident vertebrate community in the high arctic. – Science 339: 313–315. Google Scholar

30.

Henriksen, S. et al. 2003. Does availability of resources influence grazing strategies in female Svalbard reindeer? – Rangifer 23: 25–37. Google Scholar

31.

Johansen, B. and Tømmervik, H. 2014. The relationship between phytomass, NDVI and vegetation communities on Svalbard. – Int. J. Appl. Earth Observ. Geoinform. 27: 20–30. Google Scholar

32.

Johansen, B. E. et al. 2012. Vegetation mapping of Svalbard utilising Landsat TM/ETM+ data. – Polar Record 48: 47–63. Google Scholar

33.

Johnson, D. H. 1980. The comparison of usage and availability measurements for evaluating resource preference. – Ecology 61: 65–71. Google Scholar

34.

Jones, E. S. et al. 2006. Comparison of seasonal habitat selection between threatened woodland caribou ecotypes in central British Columbia. – Rangifer 17: 111–128. Google Scholar

35.

Karlsen, S. R. et al. 2014. Spatial and temporal variability in the onset of the growing season on Svalbard, Arctic Norway – measured by MODIS-NDVI satellite data. – Remote Sens.-Basel 6: 8088–8106. Google Scholar

36.

Kavan, J. 2018. Observation of polar bear Ursus maritimus feeding on Svalbard reindeer Rangifer tarandus platyrhyncus – exceptional behaviour or upcoming trend. – Czech Polar Rep. 8: 243–248. Google Scholar

37.

Kelleyhouse, R. A. 2001. Calving ground habitat selection: Teshekpuk Lake and Western Arctic caribou herds. – MS thesis, Univ. of Alaska, Fairbanks, AK. Google Scholar

38.

Klaczek, M. R. 2015. Denning ecology of barren-ground wolves in the central Candian Arctic. – MSc thesis. Univ. of Northern British Columbia, Prince George. Google Scholar

39.

Kohler, J. and Aanes, R. 2004. Effect of winter snow and ground-icing on a svalbard reindeer population: results of a simple snowpack model. – Arct. Antarct. Alp. Res. 36: 333–341. Google Scholar

40.

Lafontaine, A. et al. 2017. Many places called home: the adaptive value of seasonal adjustments in range fidelity. – J. Anim. Ecol. 86: 624–633. Google Scholar

41.

Loe, L. E. et al. 2006. Testing five hypotheses of sexual segregation in an arctic ungulate. – J. Anim. Ecol. 75: 485–496. Google Scholar

42.

Loe, L. E. et al. 2016. Behavioral buffering of extreme weather events in a high-Arctic herbivore. – Ecosphere 7: e01374. Google Scholar

43.

Mahoney, S. P. and Schaefer, J. A. 2002. Long-term changes in demography and migration of newfoundland caribou. – J. Mammal. 83: 957–963. Google Scholar

44.

Manly, B. F. J. et al. 2002. Resource selection by animals: statistical design and analysis for field studies. – Kluwer Academic Publishers. Google Scholar

45.

Mårell, A. and L. Edenius. 2006. Spatial heterogeneity and hierarchical feeding habitat selection by Reindeer. – Arct. Antarct. Alp. Res. 38: 413–420. Google Scholar

46.

Markussen, K. A. 1983. Temperatur-regulering og klimatoleranse hos nyfødte kalver av norsk rein, svalbardrein og moskus. – MSc UiT – Arctic Univ. of Norway, Tromsø. Google Scholar

47.

Mayor, S. J. et al. 2009. Habitat selection at multiple scales. – Ecoscience 16: 238–247. Google Scholar

48.

McNamara, J. M. and Houston, A. I. 1986. The common currency for behavioral decisions. – Am. Nat. 127: 358–378. Google Scholar

49.

Mettke-Hofmann, C. 2017. Avian movements in a modern world: cognitive challenges. – Anim. Cogn. 20: 77–86. Google Scholar

50.

Moullec, M. L. et al. 2017. Ungulate population monitoring in an open tundra landscape: distance sampling versus total counts. – Wildl. Biol. 2017: wlb.00299. Google Scholar

51.

Mumma, M. A. et al. 2017. Understanding predation risk and individual variation in risk avoidance for threatened boreal caribou. – Ecol. Evol. 7: 10266–10277. Google Scholar

52.

Neill, W. 1990. Induced vertical migration in copepods as a defence against invertebrate predation. – Nature 345: 524–526. Google Scholar

53.

Nelleman, C. and Thomsen, M. G. 1994. Terrain ruggedness and caribou forage availability during snowmelt on the Arctic coastal plain, Alaska. – Arctic 47: 361–367. Google Scholar

54.

Nobert, B. R. et al. 2016. Seeking sanctuary: the neonatal calving period among central mountain woodland caribou Rangifer tarandus caribou. – Can. J. Zool. 94: 837–851. Google Scholar

55.

Norwegian Polar Institute 2014. Terrengmodell Svalbard (S0 Terrengmodell) [Data set]. – Norw. Polar Inst. <  https://doi.org/10.21334/npolar.2014.dce53a47 >. Google Scholar

56.

Norwegian Meterological Inst. 2018. Varighet av snødekke på land. – Miljøovervåking Svalbard og Jan Mayen (MOSJ), <  www.mosj.no/no/ >. Google Scholar

57.

Omsjoe, E. H. et al. 2009. Evaluating capture stress and its effects on reproductive success in Svalbard reindeer. – Can. J. Zool. 87: 73–85. Google Scholar

58.

Parker, K. C. 1988. Environmental relationships and vegetation associates of columnar cacti in the northern Sonoran desert. – Vegetatio 78: 125–140. Google Scholar

59.

Pedersen, Å. ø. et al. 2017. Spatial distribution of Svalbard rock ptarmigan based on a predictive multi-scale habitat model. – Wildl. Biol. 2017: wlb.00239. Google Scholar

60.

Pedersen, S. H. et al. 2016. Spatiotemporal characteristics of seasonal snow cover in northeast greenland from in situ observations. – Arct. Antarct. Alp. Res. 48: 653–671. Google Scholar

61.

Peignier, M. et al. 2019. Space use and social association in a gregarious ungulate: testing the conspecific attraction and resource dispersion hypotheses. – Ecol. Evol. 9: 5133–5145. Google Scholar

62.

Pinard, V. et al. 2012. Calving rate, calf survival rate, and habitat selection of forest-dwelling caribou in a highly managed landscape. – J. Wildl. Manage. 76: 189–199. Google Scholar

63.

Popp, J. N. et al. 2008. Female site fidelity of the Mealy Mountain caribou herd Rangifer tarandus caribou in Labrador. – Rangifer 19: 87–95. Google Scholar

64.

Post, E. et al. 2009. Ecological dynamics across the arctic associated with recent climate change. – Science 325: 1355–1358. Google Scholar

65.

Prestrud, P. 1992. Food habits and observations of the hunting behaviour of Arctic foxes, Alopex lagopus, in Svalbard. – Can. Field-Nat. 106: 225–236. Google Scholar

66.

Pyare, S. and Berger, J. 2003. Beyond demography and delisting: ecological recovery for Yellowstone's grizzly bears and wolves. – Biol. Conserv. 113: 63–73. Google Scholar

67.

Qin, X. B. 2011. Summer bed-site selection by roe deer in a predator free area. – Hystrix 22: 269–279 Google Scholar

68.

Recio, M. R. et al. 2011. Lightweight GPS-tags, one giant leap for wildlife tracking? An assessment approach. – PLoS One 6: e28225. Google Scholar

69.

Refsnider, J. M. and Janzen, F. J. 2010. Putting eggs in one basket: ecological and evolutionary hypotheses for variation in oviposition-site choice. – Annu. Rev. Ecol. Evol. Syst. 41: 39–57. Google Scholar

70.

Rode, K. D. et al. 2015. Can polar bears use terrestrial foods to offset lost ice-based hunting opportunities? – Front. Ecol. Environ. 13: 138–145. Google Scholar

71.

Ruckstuhl, K. E. and Neuhaus, P. 2002. Sexual segregation in ungulates: a comparative test of three hypotheses. – Biol. Rev. 77: 77–96. Google Scholar

72.

Rudolph, T. and Drapeau, P. 2010. Using movement behaviour to define biological seasons for woodland caribou. – Rangifer 20: 295–307. Google Scholar

73.

Russell, D. E. et al. 2002. Barren-ground caribou calving ground workshop report of proceedings. Tech. Rep. Ser. No. 390. – Can. Wildl. Service. Google Scholar

74.

Sappington, J. M. et al. 2007. Quantifying landscape ruggedness for animal habitat analysis: a case study using bighorn sheep in the mojave desert. – J. Wildl. Manage. 71: 1419–1426. Google Scholar

75.

Schaefer, J. A. et al. 2000. Site fidelity of female caribou at multiple spatial scales. – Landscape Ecol. 15: 731–739. Google Scholar

76.

Schmidt, N. M. et al. 2019. An ecosystem-wide reproductive failure with more snow in the Arctic. – PLoS Biol. 17: e3000392. Google Scholar

77.

Stempniewicz, L. et al. 2013. Unusual hunting and feeding behaviour of polar bears on Spitsbergen. – Polar Record 50: 216–219. Google Scholar

78.

Switzer, P. V. 1993. Site fidelity in predictable and unpredictable habitats. – Evol. Ecol. 7: 533–555. Google Scholar

79.

Tremblay, J. P. et al. 2007. Fidelity to calving areas in moose Alces alces in the absence of natural predators. – Can. J. Zool. 85: 902–908. Google Scholar

80.

Tveraa, T. et al. 2013. Population densities, vegetation green-up and plant productivity: impacts on reproductive success and juvenile body mass in reindeer. – PLoS One 8: e56450. Google Scholar

81.

Tyler, N. J. C. 1986. Reinen i Adventdalen. – In: øritsland, N. A. (ed.), Svalbadreinen og dens livsgrunnlag. Universitetsforlaget, Oslo, in Norwegian. Google Scholar

82.

Tyler, N. J. C. 1987. Natural limitation of the abundance of the high arctic Svalbard reindeer. – PhD thesis, Univ. of Cambridge. Google Scholar

83.

Tyler, N. J. C. and øritsland, N. A. 1989. Why don't Svalbard reindeer migrate. – Holarctic Ecol. 12: 369–376. Google Scholar

84.

Tyler, N. J. C. et al. 2008. Nonlinear effects of climate and density in the dynamics of a fluctuating population of reindeer. – Ecology 89: 1675–1686. Google Scholar

85.

Viejou, R. et al. 2018. Woodland caribou habitat selection patterns in relation to predation risk and forage abundance depend on reproductive state. – Ecol. Evol. 8: 5863–5872. Google Scholar

86.

Wolcott, D. M. et al. 2015. Biological and environmental influences on parturition date and birth mass of a seasonal breeder. – PLoS One 10: e0124431. Google Scholar
© 2021 The Authors. This is an Open Access article This work is licensed under the terms of a Creative Commons Attribution 4.0 International License (CC-BY). The license permits use, distribution and reproduction in any medium, provided the original work is properly cited.
Ingrid M. Garfelt-Paulsen, Eeva M. Soininen, Virve Ravolainen, Leif Egil Loe, Brage B. Hansen, R. Justin Irvine, Audun Stien, Erik Ropstad, Vebjørn Veiberg, Eva Fuglei, and Åshild Ønvik Pedersen "Don't go chasing the ghosts of the past: habitat selection and site fidelity during calving in an Arctic ungulate," Wildlife Biology 2021(2), wlb.00740-, (1 April 2021). https://doi.org/10.2981/wlb.00740
Accepted: 17 February 2021; Published: 1 April 2021
JOURNAL ARTICLE
PAGES


SHARE
ARTICLE IMPACT
Back to Top