Skip to main content

Table 2 Overview of covariates considered in the HMMs for the snow-free summer and snow-covered winter bursts

From: An application of upscaled optimal foraging theory using hidden Markov modelling: year-round behavioural variation in a large arctic herbivore

Covariate type

Covariate

Description

Biological effect

Data type

Spatial/temporal resolution

Data source

temporal

time of day

hour of the day

diel variation in environmental conditions, associated with predation risk levels

continuous

hourly

 
 

Julian day

day of the year

proxy for fine-scale seasonal variation in environmental conditions and diet quality

continuous

daily

 

year

season-year (e.g. winter season 2013/2014, summer season 2014)

interannual variation in environmental conditions

categorical

annual

 

light

light conditions (daylight or darkness) at time of observation

light, visibility, associated with predation risk levels

categorical

hourly

determined using ‘streamMetabolism’ package in R

static

landcover type

NDVI-derived landcover classification (NDVI ≥0.35 = ‘dense vegetation’, 0.1–0.35 = ‘sparse vegetation’, <  0.1 = ‘bare ground’ (including non-vegetated areas such as glaciers, perennial snow and lakes))

associated with plant productivity, forage abundance

categorical

30 m

vegetation classes classified based on NDVI, using Landsat 4-5TM satellite image, dated 17 July 2009; non-vegetated derived from 1:100.000 topographic maps, field measurements from study area [23, 33]

 

elevation (m.a.s.l.)

elevation above sea level

associated with plant productivity and snow accumulation

continuous

30 m

ASTER Global Digital Elevation Model (DEM) Version 2 (https://asterweb.jpl.nasa.gov/gdem.asp)

terrain ruggedness (index)

mean of the absolute differences between the value of a cell and the value of its 8 surrounding cells, i.e. measure of terrain heterogeneity

associated with vegetation heterogeneity and variation in snow conditions

continuous

30 m

calculated from DEM using ‘terrain’ function in ‘raster’ package in R

distance to coast (m)

Euclidian distance to coastline

proxy for coast-inland gradients in e.g. precipitation, temperature

continuous

30 m

calculated from DEM using ‘raster’ package in R

hillshade (unitless)

amount of incoming radiation, combining slope and aspect

associated with local temperature, plant productivity and snow melt dynamics

continuous

30 m

calculated from DEM using ‘hillShade’ function in ‘raster’ package in R

dynamic

snow depth (m)

snow depth

associated with forage accessibility and costs of foraging/movement

continuous, modelled

300 m, 3 h

MicroMet high-resolution meteorological model coupled with SnowModel snow-evolution modelling tool [17, 18, 34]

 

ambient temperature (°C)

ambient air temperature (2 m above ground surface)

thermal conditions, associated with insect harassment

   

wind speed (m/s)

wind speed (2 m above ground surface)

associated with thermal conditions (windchill effect) and insect harassment

   

wind direction (degrees from north)

wind direction (2 m above ground surface)

associated with thermal conditions (windchill effect)

   

precipitation (mm)

precipitation (rainfall or snow) at time t

precipitation, associated with thermal conditions

   

NDVI (index)

Normalized Difference Vegetation Index (NDVI)

measure of vegetation greenness, related to vegetation growth and aboveground biomass [35]

continuous, observed

300 m, daily

Moderate Resolution Imaging Spectroradiometer (MODIS) Daily Surface Reflectance [34]