BIOL 4150 Lecture Notes - Lecture 7: Brownian Bridge, Ungulate, Lichen
Chapter 3: pages 35-52
10/31/17
Area used during the normal range of activities
•
Can be measured over lifetime or on annual, season
and daily basis
•
Satellite collars
○
Usually measured from direct observations or from
radio-telemetry fixes
•
Minimum convex polygon
!
Fixed kernel
!
Brownian bridge kernel
!
Locally-weighted polygon
!
This information can be translated into a home
range estimate
○
Ex. Daily satellite radio-telemetry fixes for gray wolf
in northern Ontario
•
Home Range:
Preferential use of particular habitat types
•
Selection ratio (use/availability)
○
Composition analysis -ln(use of i/use of
reference)
○
Relative utilization frequency
○
Resource selection function
○
It can be measured in many ways:
•
To assess the critical habitat requirements of a
species
○
To predict distribution patterns for a
population
○
Why study habitat selection?
•
Habitat Selection:
Sample of local habitat fragments are characterized
as being used (1) or unused (0)
•
∏i = the proportionate area of habitat I (ai) relative
to total area (at)
•
Oi = proportion of occurences of use of habitat 'i'
relative to total occurences
•
Selection ratio: Wi = Oi / ∏i
•
*see equation on slide
○
If unsure about sites not used, use logistic regression
to estimate probability of use
•
In this case, the predictions are not true
probabilities but are scaled to probabilities of
use
○
Demonstrates sigmoidal curve
○
Absence of data just means no evidence
○
No longer treat it as a probability
○
*see equation on slide
○
Large beta --> stronger relationship
!
Beta0 = intercept
!
Parameters: beta (strength of curvature)
○
Ex. Probability of use vs. Density of
Humans
!
*could include both factors
!
Ex. Probability of use vs. food availability
○
If unsure about sites not ever being used, use log-
linear regression model
•
Binary data vs. logistic regression --> see slide
•
Resource Selection Functions:
Location scored based on a variety of
attributes
○
Un-weighted wolf use (in Winter)
•
Potential kill site: at least 3 wolf locations
within <175m radius and >10 hours
○
Activity data from collars greatly improved
out ability to remotely detect kill sites and
reduce visits to non-feeding sites
○
Correct assignment 92% of the time
○
Results: moose*, caribou, bears, unknown
ungulate
○
Kill rates:
•
Do not need to move far to acquire food
!
More competition (due to more prey
resources)
!
High density of prey --> smaller territory
○
Dumps: conservation choice
□
Roads
□
May include:
!
*focus conserved and managed areas to ensure
viability of moose populations
○
Wolf home range size and moose density:
•
Ex. Wolves in Northern Ontario
Have things in common (ex. Predator)
○
Increase in moose --> increase predation
of wolves on moose
!
Causes wolves population to increase
and augments predation risk for caribou
!
Therefore, moose increase predation risk
for caribou
!
Competition is mediated by the predator
○
There is apparent competition between wolves,
caribou and moose
•
Linear features (km/km^2): 0.0423 vs. 0.4755
○
Moose (/100km^2): 2.4 vs. 4.6
○
% Area burned: 12.98 vs. 4.94
○
% Coniferous: 41.98 vs. 29.17
○
Wolves (/100km^2): 0.3 vs. 0.5
○
Pickle lake (control) & Auden sites (disturbed
treatment)
•
34 individuals
○
51-289 days / individual
○
Acceleration was sampled
!
X and Y axis
!
…
!
Collars:
○
Information was gathered on caribou via capture and
video collars
•
Majority is lichen
○
Caribou diets by season and site
•
Over 200 stands were sampled for vegetation
abundance
•
Can determine energetic costs of each
behaviour
○
Activity levels correspond to behaviour
•
--> wolves & food availability attributes
○
*see digestible food biomass & wolf density
vs. frequency (used vs unused)
○
Not perfect avoidance or overlap
○
Compared spatial distribution of wolves
•
Mortality is majorly due to disturbed predation
•
Leslie Matrix Model
•
Anthropogenic disturbance and population viability of
woodland caribou:
Home Range and Habitat Use
Tuesday,+ October+31,+2017
11:28+AM
Document Summary
Area used during the normal range of activities. Can be measured over lifetime or on annual, season and daily basis. Usually measured from direct observations or from radio-telemetry fixes. Daily satellite radio-telemetry fixes for gray wolf in northern ontario. This information can be translated into a home range estimate. Composition analysis - ln(use of i/use of reference) To assess the critical habitat requirements of a species. Sample of local habitat fragments are characterized as being used (1) or unused (0) I = the proportionate area of habitat i (ai) relative to total area (at) Oi = proportion of occurences of use of habitat "i" relative to total occurences. If unsure about sites not used, use logistic regression to estimate probability of use. If unsure about sites not ever being used, use log- linear regression model. In this case, the predictions are not true probabilities but are scaled to probabilities of use. Binary data vs. logistic regression --> see slide.