Materials and methods:
STUDY AREA
The study area encompassed the winter and summer ranges of the partially
migratory Ya Ha Tinda elk herd along the eastern slopes of the Rocky
Mountains in and adjacent to Banff National Park (BNP, Fig. 1).
High-elevation bare rock and mixed shrub and herbaceous alpine
communities dominated areas > 2,100 m in the west.
Engelmann spruce (Picea engelmannii ) and subalpine fir
(Abies lasiocarpa ) were the primary high-elevation conifer
landcover, with low-elevation forests consisting of lodgepole pine
(Pinus contorta ) and white spruce (P. glauca ). Early seral
stands (< 20-year stand age) consisted of logged areas
(hereafter, “cutblocks”) and post-fire forest regeneration.
Other ungulates in the area include white-tailed deer (Odocoileus
virginianus ), mule deer (O. hemionus ), moose, bighorn sheep
(Ovis canadensis ), mountain goats (Oreamnos americanus ),
and feral horses (Equus caballus ). Wolves naturally recolonized
the study in the mid-1980s and continue to be relatively stable (Berg,
2019). Grizzly bears have increased in Alberta (Morehouse & Boyce,
2016), and densities on protected federal land were 2.4 times higher
than on provincial lands (Boulanger et al., 2005; Whittington & Sawaya,
2015). Cougars expanded their range in northern and eastern Alberta
since the 1990s (Knopff et al., 2014).
SCAT COLLECTION AND ANALYSIS
We used scat-detection dogs to detect predator scats during 1 July – 30
September, 2013 – 2016, along stratified random transects located in
proportion to elevation classes and landcover representation within a
systematic sampling grid of 57 5 x 5-km cells (see Spilker, 2019). We
recorded scat diameter and physical description to identify scats to
species (Weaver & Fritts, 1979; Rezendes, 1992; Elbroch, 2003), and
collected DNA on a subsample of scats to assess our accuracy. We
combined grizzly and black bears into one ursid category because we
found low accuracy in our ability to discriminate the two (Spilker,
2019).
We analyzed scat contents for the presence of elk hair using either
macroscopic analysis or DNA analysis. For macroscopic analysis, we
randomly selected 20 hairs from each scat, prepared hairs using standard
methods (Ciucci et al., 1996), and identified the species based on
characteristics of the hairs’ medulla, cuticle scale patterns, and scale
margin distance using dichotomous keys (Moore et al., 1974; Kennedy &
Carbyn, 1981). Three trained observers were subject to blind trials on
known hairs, obtaining a minimum of 80% correct classification rate
prior to analysis.
DNA was extracted from hair shafts using QIAGEN’s DNeasy Tissue kits
(QIAGEN Inc., Valencia, USA). Polymerase chain reaction (PCR) was used
to amplify DNA and prey species identification was confirmed via a
partial sequence analysis of a hypervariable region of the mitochondrial
16S rRNA gene. This approach identified the most dominant prey species
in the scat (i.e., based on the proportion of DNA); mixed samples
where there was no dominant species (or equal amounts of DNA from each
species) were re-run with ungulate-specific primers to determine if elk
DNA was present. We compared the presence of elk from the DNA analysis
to the macroscopic analysis on the same scats (n = 60) based on
Area Under the Curve from a Receiver Operating Characteristic curve. We
found DNA analysis detected elk present in 88% of the scats where we
detected elk macroscopically. Wildlife Genetics International (Nelson,
Canada) performed DNA analyses.
SPATIAL PREDATION RISK
Spatial predation risk (PRij ) reflected the
relative risk of an elk dying from a specific predator, i , at a
location j , and was derived as:
PRij = RSFij *
Pij eqn 2
where RSFij is scat-based resource selection
function, and Pij is relative probability of elk
being in the predator scat at the location. To estimate total risk from
all predators (bear, cougar, coyote, or wolf), we summed the individual
predation risk predictions standardized from (0 – 1).