Reindeer movement, habitat preference and road permeability model data

Full description

Reindeer movement, habitat preference and road permeability model data Overview: GPS data for wild reindeer were collected within a larger project in Rondane-South and Rondane-North wild reindeer management areas, a mountainous region of central-southern Norway (10 46’ E, 61 38’ N). We used locations collected from five adult female reindeer every three hours between 1 June to 29 September 2012 (N = 973, 960, 871, 971 and 974 locations, respectively). Around 60% of the area is located above tree-line between 1000 and 1500 m, and is dominated by rocks and lichen heath; lower elevations (above 500 m) are characterised by a mix of meadows, grass and willow communities. The area occupied by the reindeer used in this study extends between ca. 400 and 1900 m, and is fragmented by public and private roads (access to the latter is often restricted, so is characterised by lower traffic volumes than the former). The data we provide can be used in conjunction with the R code in the Supplementary Materials of the published paper to fit the models presented in that paper. The zip file contains five R data files corresponding to five reindeer. The files are named "data_" followed by one of the ID numbers of the reindeer (11264, 11265, 9397, 7625, 9406). Each R data (*.RData) file contains an R list object called "fitdata" in R, composed of the following data structures: Definitions: N_s = the number of movement "steps" (sequential telemetry locations) for a reindeer N_a = the number of grid cells in the landscape (all cells falling within 5km of any telemetry point. (i) fitdata$usehab A matrix (dimensions: N_s rows, 4 columns) of 'used' habitat types (the habitat value at each telemetry location in the movement path). The four columns are: elevation (km), elevation^2, distance to road (km), distance to road^2. The squared terms are included because the habitat preference model uses quadratic terms to allow for non-linear preference with respect to the two habitat covariates. (ii) fitdata$availhab A matrix (dimensions: N_a rows, 4 columns) of 'available' habitat types (the habitat value at each raster grid cell in the spatial domain of the analysis). The four columns are: elevation (km), elevation^2, distance to road (km), distance to road^2. (iii) fitdata$use.xy A matrix (dimensions N_s + 1 rows, 2 columns) of the x and y coordinates (km) of the reindeer telemetry locations. This matrix is used to precalculate a large distance matrix ("dm") representing the distances among all use and available points (dimensions: N_s rows, N_a columns). Although this 'dm' matrix is very large, pre-calculating the distance matrix greatly improves the speed of model fitting by eliminating the need to repeatedly re-calculate distances. (iv) fitdata$stepdst A vector of length N_s - 1 representing the Euclidean distance (km) between consecutive telemetry locations. (v) fitdata$avail.xy A matrix (dimensions N_a rows, 2 columns) of the x and y coordinates (km) of the grid cells forming the 'availability' sample. (vi) fitdata$use.rdzn A vector of length N_s containing arbitrary ID numbers corresponding to regions of space that are bounded by a network of roads. This could be conceptualised in GIS terms as forming polygons based on the lines contained in the road network, and assigning an arbitrary unique ID to each of those polygons. Road crossings are indicated by a change in the ID number in this time series. The reason for using this approach to identifying road crossings is computational efficiency, and to avoid the assumption that the straight-line connected two consecutive telemetry locations can be used to calculate road crossings. For example, if this straight line 'clips' a bend in a road then that would indicate 2 crossings, while the animal may in fact have remained in the same region without ever crossing a road. (vii) fitdata$avail.rdzn A vector of length N_a containing arbitrary ID numbers corresponding to regions of space that are bounded by a network of roads (see vi) for each grid cell in the availability sample. Telemetry and habitat data for 5 reindeer over 1 summer; R data objects.