Data from: Delineating seasonal shifts in Reindeer habitat and diet selection by integrating GPS telemetry and stable Isotope analysis
Data files
Dec 30, 2025 version files 8.21 MB
-
GPS_Data.csv
293.54 KB
-
GPS_DATA.RMD
172.35 KB
-
Habitat_Diet_LMM.Rmd
35.73 KB
-
ISOTOPE_DATA.Rmd
14.49 KB
-
Kaldoaiviboundary2.shp-polygon.cpg
5 B
-
Kaldoaiviboundary2.shp-polygon.dbf
76 B
-
Kaldoaiviboundary2.shp-polygon.prj
143 B
-
Kaldoaiviboundary2.shp-polygon.shp
14.12 KB
-
Kaldoaiviboundary2.shp-polygon.shx
108 B
-
README.md
21.19 KB
-
RSF_Data.csv
7.64 MB
-
RSF_MODEL_OUTPUTS.csv
7.16 KB
-
RSF_SIA_TRAJ.csv
4.16 KB
-
RSF_SIA.csv
2.94 KB
Abstract
Seasonal changes shape herbivore behaviour by altering forage availability and habitat conditions; however, few studies integrate diet and habitat selection data across temporal scales. This study uses seasonality as a unifying framework to combine fine-scale GPS-based habitat selection data with broader-scale dietary information from stable isotope analysis (δ¹³C, δ¹⁵N) of hairs in semi-domesticated reindeer (Rangifer tarandus tarandus). Thresholds introduced within this framework detect seasonal shifts in habitat and diet selection and classify foraging strategies along a specialist-generalist continuum. Despite individual variability, most individuals exhibited pronounced seasonal changes between spring/early summer (SES) and late summer/autumn (LSA), consistent with generalist foraging strategies. Habitat selection models revealed reduced avoidance of rugged terrain and increased use of mires and bogs from SES to LSA. Concurrently, isotopic enrichment and niche expansion reflected a dietary shift from δ¹⁵N- and δ¹³C-depleted plants (e.g., lichens, shrubs, deciduous vegetation) to more enriched forage types such as sedges, grasses, horsetails, and mushrooms. We assessed whether individual-level shifts in diet and habitat use were linked and found an inverse correlation between the shifts in terrain ruggedness avoidance and dietary change in approximately 67 % of individuals, suggesting that behavioural flexibility facilitates seasonal transitions. By integrating spatial and isotopic data, this study overcomes the limitations of single-method approaches and provides a more nuanced understanding of the seasonal foraging dynamics of a keystone Arctic and boreal ungulate. The findings highlight the plasticity of reindeer foraging within a variable environment and suggest a capacity to respond to environmental changes. This framework also offers broader applications for investigating behavioural responses and ecological strategies in other herbivores facing climate-driven habitat shifts.
Dataset DOI: 10.5061/dryad.5x69p8djm
Description of the data and file structure
Reindeer collaring and hair sampling
Reindeer in Kaldoaivi are rounded up and handled for management purposes twice annually, in mid-to-late winter (February/March) and autumn (October/November), providing opportunities for collaring and sample collection. In February 2019, 15 adult female reindeer were fitted with Ultracom Poro GPS M5 collars (Ultracom Oy, Oulu, Finland) at the Skalluvaara (Skállovárri) roundup site (Figure 1c in associated publication). The GPS collars recorded locations at a frequency of four locations per day. Collars were removed during autumn 2019 (N = 13) and spring 2020 (N = 2) roundups, and hair samples were collected from the rump next to the tail and stored in paper bags (Mosbacher et al., 2016). All collars were sent to Ultracom Oy (Oulu, Finland) for the data to be downloaded.
Sample preparation and stable isotope analysis
Reindeer hair samples were processed for stable isotope analysis (SIA) following established protocols in the literature (Hiltunen et al., 2022; Massé et al., 2023; Rogers et al., 2020). Samples were washed overnight in 2:1 chloroform/methanol (67/33, v/v) on an orbital shaker (180 oscillations/minute) to remove sebum lipids. They were then rinsed with fresh solvent, ethanol (twice), and distilled water (twice) and subsequently dried at 50 °C for 24 h. The longest whole hairs (15-20/sample) were cut into 2 cm sections using a surgical scalpel. Of 13 samples, only 12 had both follicle (A) and tip (B) sections, and the hair sections of each reindeer were pooled accordingly. Hair sections (0.8 to 1.2 mg) were weighed into 5 × 9 mm tin capsules (Elemental Microanalysis Ltd, UK).
Samples were analysed at the Environment and Natural Resources Institute Stable Isotope Laboratory at the University of Alaska Anchorage (http://www.uaa.alaska.edu/enri/labs/sils). δ13C and δ15N values (‰) were calibrated against internal standards (BWBII keratin, peach leaves, moose [Alces alces], and three-spined stickleback [Gasterosteus aculeatus]) and referenced to the international standards of Vienna Pee Dee Belemnite (VPDB) for δ13C and atmospheric N₂ (air) for δ15N. Concentrations of the heavy isotopes (13C and 15N) are reported using the standard delta notation (δ) in parts per thousand relative to an international standard. The sample analytical error was ± 0.1 ‰ for both C and N.
Files and variables
File: GPS_Data.csv
Description: GPS data from the collared reindeer
Variables
- ID: Reindeer ID
- Easting: GPS coordinate UTM Easting
- Northing: GPS coordinate UTM Northing
- Season: Seasons in study 2 = Spring/ Early Summer (SES); 3 = Late Summer/Autumn (LSA)
- TimeStamp: Time stamp in DD/MM/YYYY HH:MM:SS
- Lon: GPS coordinate Decimal Degrees Longitude (WGS84)
- Lat: GPS coordinate Decimal Degrees Latitude (WGS84)
File: GPS_DATA.RMD
Description: R-markdown script for the preparation of the GPS data for analysis and resource selection function models
File: Kaldoaiviboundary2.shp-polygon.cpg
Description: File associated with the Shapefile polygon of the Kaldoaivi Reindeer Herding District Boundary
File: Kaldoaiviboundary2.shp-polygon.dbf
Description: File associated with the Shapefile polygon of the Kaldoaivi Reindeer Herding District Boundary
File: Kaldoaiviboundary2.shp-polygon.prj
Description: File associated with the Shapefile polygon of the Kaldoaivi Reindeer Herding District Boundary
File: Kaldoaiviboundary2.shp-polygon.shp
Description: Shapefile polygon of the Kaldoaivi Reindeer Herding District Boundary
File: Kaldoaiviboundary2.shp-polygon.shx
Description: File associated with the Shapefile polygon of the Kaldoaivi Reindeer Herding District Boundary
File: RSF_MODEL_OUTPUTS.csv
Description: Model outputs after the GPS data has been cleaned, random points have also been generated and Resource Selection function models have been run for each individual reindeer and season.
Variables
- ID: Reindeer ID
- Season: SES = Spring/ Early Summer ; LSA = Late Summer/Autumn
- No_Rel: the no of GPS locations for the reindeer
- HR_KM: Size of home range calculated for each reindeer for each season during the RSF modelling process
- Parameter: Name of the model output term:
StdError = The standard error associated with the coefficient estimate.;
Estimate = The coefficient estimate from the RSF model. This is the estimated effect of the parameter on the log-odds of selection.;
OddsRatio = Exponentiated coefficient (exp(Estimate)), representing the multiplicative change in selection odds for a one-unit increase in the covariate.
OddsRatioSE = The standard error of the odds ratio, derived from the coefficient’s standard error. - Intercept: the model’s constant term for that variable
- elev: elevation variable - values derived from the models for each parameter
- rug: Terrain ruggedness variable - values derived from the models for each parameter
- NDVI: NDVI variable - values derived from the models for each parameter
- DSGP: pasture class variable Dwarf Shrub and Grass Pasture relative to Lichen pasture - values derived from the models for each parameter
- ALP: pasture class variable Arboreal Lichen Pasture relative to Lichen pasture - values derived from the models for each parameter
- WS: pasture class variable Water System relative to Lichen pasture - values derived from the models for each parameter
- MB: pasture class variable Mires and Bogs relative to Lichen pasture - values derived from the models for each parameter
- BFM: pasture class variable Bare Fell and Mountain Pasture relative to Lichen pasture - values derived from the models for each parameter
File: RSF_Data.csv
Description: Output after the GPS data has been cleaned and random points have also been generated and the data extracted from all the variable raster files for each individual reindeer and season.
Variables
- case_: TRUE = Used or actual reindeer GPS locations ; FALSE = Avaialble but not used randomly selected locations
- x_: GPS coordinate UTM in meters Easting
- y_: GPS coordinate UTM in meters Northing
- ID: Reindeer ID
- Season: Seasons in study 2 = Spring/ Early Summer (SES); 3 = Late Summer/Autumn (LSA); ALL = randomly selected data points not related to season (used in analyses for SES and LSA)
- elev: extracted scaled elevation covariate (DEM) at the location (scaled where mean is 0 and the standard deviation is 1)
- past: extracted pasture class covariate at the location: Pasture classes = Lichen pasture, Dwarf shrub and grass pasture, Arboreal lichen pasture, Arable land, sand and gravel, Infrastructure area, Water system, Agricultural area, Mires and bogs, Bare fjeld / mountain"
- rug: extracted scaled errain ruggedness covariate (derived from the DEM) at the location (scaled where mean is 0 and the standard deviation is 1)
- NDVI: extracted NDVI covariate at the location. Where the NDVI value is divided by 1000 and tehn scaled, where mean is 0 and the standard deviation is 1)
File: ISOTOPE_DATA.Rmd
Description: R-markdown script for the preparation of the Isotope data for analysis and calculation of the isotopic trajectories and combining with estimates from the resource selection function models.
File: RSF_SIA.csv
Description: File containing both the isotope data and the outputs from the resource selection function models
Variables
- ID: Reindeer ID
- Reindeer_No: Number of the reindeer extracted from the ID (e.g. K1 = 1)
- d13C_SES: δ13C value for hair section from Spring/ Early Summer (SES)
- d15N_SES: δ15N value for hair section from Spring/ Early Summer (SES)
- d13C_LSA: δ13C value for hair section from Late Summer/Autumn (LSA)
- d15N_LSA: δ15N value for hair section from Late Summer/Autumn (LSA)
- n_SES: number of GPS locations collected by per reindeer in Spring/ Early Summer (SES)
- n_LSA: number of GPS locations collected by per reindeer in Late Summer/Autumn (LSA)
- HR_LSA: home range size in km2 in Late Summer/Autumn (LSA)
- HR_SES: home range size in km2 in Spring/ Early Summer (SES)
- elev_B_LSA: elevation selection coefficient (β) from resource selection function models for each individual in Late Summer/Autumn (LSA)
- rug_B_LSA: terrain ruggedness selection coefficient (β) from resource selection function models for each individual in Late Summer/Autumn (LSA)
- NDVI_B_LSA: NDVI selection coefficient (β) from resource selection function models for each individual in Late Summer/Autumn (LSA)
- DSGP_B_LSA: dwarf shrub and grass pasture class (DSGP) relative to lichen pasture (LP) selection coefficient (β) from resource selection function models for each individual in Late Summer/Autumn (LSA)
- MB_B_LSA: mires and bogs pasture class (MB) relative to lichen pasture (LP) selection coefficient (β) from resource selection function models for each individual in Late Summer/Autumn (LSA)
- BFM_B_LSA: bare fell and mountain pasture class (BFM) relative to lichen pasture (LP) selection coefficient (β) from resource selection function models for each individual in Late Summer/Autumn (LSA)
- BFM_prop_LSA: proportion of bare fell and mountain pasture (BFM) class used in Late Summer/Autumn (LSA)
- DSGP_prop_LSA: proportion of dwarf shrub and grass pasture (DSGP) class used in Late Summer/Autumn (LSA)
- LP_prop_LSA: proportion of lichen pasture (LP) class used in Late Summer/Autumn (LSA)
- MB_prop_LSA: proportion of mires and bogs pasture (MB) class used in Late Summer/Autumn (LSA)
- elev_SE_LSA: elevation selection coefficient (β) standard error (SE) from resource selection function models for each individual in Late Summer/Autumn (LSA)rug_SE_LSA: Late Summer/Autumn (LSA)
- NDVI_SE_LSA: NDVI selection coefficient (β) standard error (SE) from resource selection function models for each individual in Late Summer/Autumn (LSA)
- DSGP_SE_LSA: dwarf shrub and grass pasture class (DSGP) relative to lichen pasture (LP) selection coefficient (β) standard error (SE) from resource selection function models for each individual in Late Summer/Autumn (LSA)
- MB_SE_LSA: mires and bogs pasture class (MB) relative to lichen pasture (LP) selection coefficient (β) standard error (SE) from resource selection function models for each individual in Late Summer/Autumn (LSA)
- BFM_SE_LSA: bare fell and mountain pasture class (BFM) relative to lichen pasture (LP) selection coefficient (β) standard error (SE) from resource selection function models for each individual in Late Summer/Autumn (LSA)
- elev_B_SES: elevation selection coefficient (β) from resource selection function models for each individual in Spring/ Early Summer (SES)
- rug_B_SES: terrain ruggedness selection coefficient (β) from resource selection function models for each individual in Spring/ Early Summer (SES)
- NDVI_B_SES: NDVI selection coefficient (β) from resource selection function models for each individual in Spring/ Early Summer (SES)
- DSGP_B_SES: dwarf shrub and grass pasture class (DSGP) relative to lichen pasture (LP) selection coefficient (β) from resource selection function models for each individual in Spring/ Early Summer (SES)
- MB_B_SES: mires and bogs pasture class (MB) relative to lichen pasture (LP) selection coefficient (β) from resource selection function models for each individual in Spring/ Early Summer (SES)
- BFM_B_SES: bare fell and mountain pasture class (BFM) relative to lichen pasture (LP) selection coefficient (β) from resource selection function models for each individual in Spring/ Early Summer (SES)
- BFM_prop_SES: proportion of bare fell and mountain pasture (BFM) class used in Spring/ Early Summer (SES)
- DSGP_prop_SES: proportion of dwarf shrub and grass pasture (DSGP) class used in Spring/ Early Summer (SES)
- LP_prop_SES: proportion of lichen pasture (LP) class used in Spring/ Early Summer (SES)
- MB_prop_SES: proportion of mires and bogs pasture (MB) class used in Spring/ Early Summer (SES)
- elev_SE_SES: elevation selection coefficient (β) standard error (SE) from resource selection function models for each individual in Spring/ Early Summer (SES)
- rug_SE_SES: terrain ruggedness selection coefficient (β) standard error (SE) from resource selection function models for each individual in Spring/ Early Summer (SES)
- NDVI_SE_SES: NDVI selection coefficient (β) standard error (SE) from resource selection function models for each individual in Spring/ Early Summer (SES)
- DSGP_SE_SES: dwarf shrub and grass pasture class (DSGP) relative to lichen pasture (LP) selection coefficient (β) standard error (SE) from resource selection function models for each individual in Spring/ Early Summer (SES)
- MB_SE_SES: mires and bogs pasture class (MB) relative to lichen pasture (LP) selection coefficient (β) standard error (SE) from resource selection function models for each individual in Spring/ Early Summer (SES)
- BFM_SE_SES: bare fell and mountain pasture class (BFM) relative to lichen pasture (LP) selection coefficient (β) standard error (SE) from resource selection function models for each individual in Spring/ Early Summer (SES)
File: Habitat_Diet_LMM.Rmd
Description: R-markdown script for the linear mixed effect models to determine if there are relationships between the isotope niche trajectories (shifts) and the shifts in the selection coefficients derived from the resource selection function models.
File: RSF_SIA_TRAJ.csv
Description: File containing both the isotope trajectories and the outputs from the resource selection function models.
Variables
-
ID: Reindeer ID
-
Reindeer_No: Number of the reindeer extracted from the ID (e.g. K1 = 1)
-
d13C_SES: δ13C value for hair section from Spring/ Early Summer (SES)
-
d15N_SES: δ15N value for hair section from Spring/ Early Summer (SES)
-
d13C_LSA: δ13C value for hair section from Late Summer/Autumn (LSA)
-
d15N_LSA: δ15N value for hair section from Late Summer/Autumn (LSA)
-
HR_LSA: home range size in km2 in Late Summer/Autumn (LSA)
-
elev_B_LSA: elevation selection coefficient (β) from resource selection function models for each individual in Late Summer/Autumn (LSA)
-
rug_B_LSA: terrain ruggedness selection coefficient (β) from resource selection function models for each individual in Late Summer/Autumn (LSA)
-
NDVI_B_LSA: NDVI selection coefficient (β) from resource selection function models for each individual in Late Summer/Autumn (LSA)
-
DSGP_B_LSA: dwarf shrub and grass pasture class (DSGP) relative to lichen pasture (LP) selection coefficient (β) from resource selection function models for each individual in Late Summer/Autumn (LSA)
-
MB_B_LSA: mires and bogs pasture class (MB) relative to lichen pasture (LP) selection coefficient (β) from resource selection function models for each individual in Late Summer/Autumn (LSA)
-
BFM_B_LSA: bare fell and mountain pasture class (BFM) relative to lichen pasture (LP) selection coefficient (β) from resource selection function models for each individual in Late Summer/Autumn (LSA)
-
BFM_prop_LSA: proportion of bare fell and mountain pasture (BFM) class used in Late Summer/Autumn (LSA)
-
DSGP_prop_LSA: proportion of dwarf shrub and grass pasture (DSGP) class used in Late Summer/Autumn (LSA)
-
LP_prop_LSA: proportion of lichen pasture (LP) class used in Late Summer/Autumn (LSA)
-
MB_prop_LSA: proportion of mires and bogs pasture (MB) class used in Late Summer/Autumn (LSA)
-
elev_SE_LSA: elevation selection coefficient (β) standard error (SE) from resource selection function models for each individual in Late Summer/Autumn (LSA)rug_SE_LSA: Late Summer/Autumn (LSA)
-
NDVI_SE_LSA: NDVI selection coefficient (β) standard error (SE) from resource selection function models for each individual in Late Summer/Autumn (LSA)
-
DSGP_SE_LSA: dwarf shrub and grass pasture class (DSGP) relative to lichen pasture (LP) selection coefficient (β) standard error (SE) from resource selection function models for each individual in Late Summer/Autumn (LSA)
-
MB_SE_LSA: mires and bogs pasture class (MB) relative to lichen pasture (LP) selection coefficient (β) standard error (SE) from resource selection function models for each individual in Late Summer/Autumn (LSA)
-
BFM_SE_LSA: bare fell and mountain pasture class (BFM) relative to lichen pasture (LP) selection coefficient (β) standard error (SE) from resource selection function models for each individual in Late Summer/Autumn (LSA)
-
HR_SES: home range size in km2 in Spring/ Early Summer (SES)
-
elev_B_SES: elevation selection coefficient (β) from resource selection function models for each individual in Spring/ Early Summer (SES)
-
rug_B_SES: terrain ruggedness selection coefficient (β) from resource selection function models for each individual in Spring/ Early Summer (SES)
-
NDVI_B_SES: NDVI selection coefficient (β) from resource selection function models for each individual in Spring/ Early Summer (SES)
-
DSGP_B_SES: dwarf shrub and grass pasture class (DSGP) relative to lichen pasture (LP) selection coefficient (β) from resource selection function models for each individual in Spring/ Early Summer (SES)
-
MB_B_SES: mires and bogs pasture class (MB) relative to lichen pasture (LP) selection coefficient (β) from resource selection function models for each individual in Spring/ Early Summer (SES)
-
BFM_B_SES: bare fell and mountain pasture class (BFM) relative to lichen pasture (LP) selection coefficient (β) from resource selection function models for each individual in Spring/ Early Summer (SES)
-
BFM_prop_SES: proportion of bare fell and mountain pasture (BFM) class used in Spring/ Early Summer (SES)
-
DSGP_prop_SES: proportion of dwarf shrub and grass pasture (DSGP) class used in Spring/ Early Summer (SES)
-
LP_prop_SES: proportion of lichen pasture (LP) class used in Spring/ Early Summer (SES)
-
MB_prop_SES: proportion of mires and bogs pasture (MB) class used in Spring/ Early Summer (SES)
-
elev_SE_SES: elevation selection coefficient (β) standard error (SE) from resource selection function models for each individual in Spring/ Early Summer (SES)
-
rug_SE_SES: terrain ruggedness selection coefficient (β) standard error (SE) from resource selection function models for each individual in Spring/ Early Summer (SES)
-
NDVI_SE_SES: NDVI selection coefficient (β) standard error (SE) from resource selection function models for each individual in Spring/ Early Summer (SES)
-
DSGP_SE_SES: dwarf shrub and grass pasture class (DSGP) relative to lichen pasture (LP) selection coefficient (β) standard error (SE) from resource selection function models for each individual in Spring/ Early Summer (SES)
-
MB_SE_SES: mires and bogs pasture class (MB) relative to lichen pasture (LP) selection coefficient (β) standard error (SE) from resource selection function models for each individual in Spring/ Early Summer (SES)
-
BFM_SE_SES: bare fell and mountain pasture class (BFM) relative to lichen pasture (LP) selection coefficient (β) standard error (SE) from resource selection function models for each individual in Spring/ Early Summer (SES)
-
trajL: niche trajectory length (i.e., magnitude of change) = Xi
-
trajA: niche trajectory direction (i.e., angle of change) = θi
-
sin_trajA: sine (sinθi) of angle of change (θi)
cos_trajA: cosine (cosθi) of angle of change (θi)
-
D13C: Change in δ13C value between SES and LSA for each individual
-
D15N: Change in δ15N value between SES and LSA for each individual
Code/software
R version 4.2.1 (R Core Team, 2022) using RStudio (version 2023.03.1+446)
R package amt (version 0.2.1.0)
R package lme4 version 1.1-32
Stable Isotope Bayesian Ellipses in R; version 2.1.6
circular package (R package version 0.4-93)
cAIC4 R package version 1.0
Access information
Data was derived from the following sources:
- NDVI values, based on red and near-infrared (NIR) reflectance ratios (Pettorelli et al., 2005) derived from the 2019 Sentinel-2 imagery, were obtained from terraPulse Inc. (North Potomac, Maryland, USA).
- Contains Elevation and terrain Ruggedness data from the Elevation model 2019, 10 m x 10 m from the National Land Survey of Finland Topographic Database 2019. CC BY 4.0.
- Vegetation coverage (30 m resolution) included nine pasture classes (Kumpula et al., 2006). Kumpula, J., Colpaert, A., Tanskanen, A., Anttonen, M., Törmänen, H., & Siitari, J. (2006). Porolaidunten inventoinnin kehittäminen - Keski-Lapin paliskuntien laiduninventointi vuosina 2005-2006 (p. 42). Kala- ja riistaraportteja 397. Riista- ja kalatalouden tutkimuslaitos.
