Institut für Statistik und Ökonometrie
20
Institut für Statistik und Ökonometrie A Wildlife Simulation Package WWalter Zucchini 1 , David Borchers 2 , Stefan Kirchfeld 1 , Martin Erdelmeier 1 Jon Bishop 2 Contributors: Shirley Pledger, Alexey Botvinnik, Günter Kratz, Oliver Kellenberger, Jürgen Meinecke Research Unit for Wildlife Population Assessment 1 University of Göttingen 2 University of St. Andrews
description
A Wildlife Simulation Package WWalter Zucchini 1 , David Borchers 2 , Stefan Kirchfeld 1 , Martin Erdelmeier 1 Jon Bishop 2. Institut für Statistik und Ökonometrie. Research Unit for Wildlife Population Assessment. 1 University of Göttingen. 2 University of St. Andrews. - PowerPoint PPT Presentation
Transcript of Institut für Statistik und Ökonometrie
Institut für Statistik und Ökonometrie
A Wildlife Simulation Package
WWalter Zucchini1, David Borchers2, Stefan Kirchfeld1, Martin Erdelmeier1 Jon Bishop2
Contributors: Shirley Pledger, Alexey Botvinnik, Günter Kratz, Oliver Kellenberger, Jürgen Meinecke
Research Unit for Wildlife Population Assessment
1 University of Göttingen 2 University of St. Andrews
Concepts in animal abundance estimation
State model Describes the spatial distribution and characteristics of animals in a region
Who lives here and where are they?
Population: group size, composition (gender, type), position, exposure
Observation modelDescribes the probability that animals with given characteristics are detected.
What we assume we are likely to see.
Detection probability: certain, constant, distance dependent, covariate-dependent
Survey designDescribes the covered region, survey units, effort (observers/traps)
Where, how, and how hard we look.
Survey design: plot sampling, removal methods, mark-recapture, distance sampling, etc.
Abundance estimation process in WiSPgenerate.region generate.density
region density1.) Define survey region & population density
functions objects
Generating a survey region
Survey region = An area of given height and width
survey region
0 20 40 60 80
0
10
20
30
40
50
Generating a population density (1)
Simple density with linear trend:
0 20 40 60 80
010
20
30
40
50
x
y
Density
high density
low density
The population density defines the spatial distribution of animals in the survey region.
Generating a population density (2)
Increase complexity by adding hotspots and coldspots:
0 20 40 60 80
010
20
30
40
50
xy
Density
hotspot
coldspot
Generating a population density (3)
Add more complexity: Strips of constant density
0 20 40 60 80
010
20
30
40
50
xy
Density
no animals here
Abundance estimation processgenerate.region generate.density
region density1.) Define survey region & population density
setpars.population
generate.population
population
2.) Generate population
functions objects
Generating a population
The population specifies the positions and characteristics of groups and individuals
Population parameters
• region
• density
• probability distributions of group and individual characteristics (group sizes and exposures)
• number of groups
Generating a population
group-size distribution
1 2 3 4 5
0.0
0.3
0.6 exposure distribution
0 20 40 60 80 100
0.00
0.06
population parameters
0 20 40 60 80
010
2030
4050
Region, density and N Generate population
Population
0 20 40 60 80
0
10
20
30
40
50
A population with 500 groups
0 20 40 60 80
0
10
20
30
40
50
large grouplow exposure
small grouphigh exposure
generate.region generate.density
region density1.) Define survey region & population density
setpars.design
generate.design
design
3.) Generate survey design
setpars.population
generate.population
population
2.) Generate population
Abundance estimation process in WiSP
functions objects
Survey designs for closed populations
• plot sampling Count all animals in a selected sub-region
• removal methodsRepeatedly remove some
• mark-recapturevarious Repeatedly
capture, mark, return
• distance samplingline transect Count along a transect – record the distancespoint transect Count within a circle – record the distances
• nearest objectsingle nearest object Measure distance to nearest object detected
Survey designs with Incomplete Coverage
0 20 40 60 80
0
10
20
30
40
50
0 20 40 60 80
0
10
20
30
40
50
plot sampling (random)
0 20 40 60 80
0
10
20
30
40
50
point transect
0 20 40 60 80
0
10
20
30
40
50
line transect
Abundance estimation process in WiSPgenerate.region generate.density
region density1.) Define survey region & population density
setpars.design
generate.design
design
3.) Generate survey design
setpars.survey
generate.sample
sample
4.) Survey population
setpars.population
generate.population
population
2.) Generate population
functions objects
0 20 40 60 80
0
10
20
30
40
50
A line-transect survey
Specify design parsGenerate design (random)
… a sample
0 20 40 60 80
0
10
20
30
40
50
0 20 40 60 80
0
10
20
30
40
50
0 20 40 60 80
0
10
20
30
40
50
Do survey to generate …
Specify survey pars
point.sim.est
simulation est
Abundance estimation process in WiSPgenerate.region generate.density
region density1.) Define survey region & population density
setpars.design
generate.design
design
3.) Generate survey design
setpars.survey
generate.sample
sample
4.) Survey population
setpars.population
generate.population
population
2.) Generate population
5.) Compute estimates
point.est int.est
point est interval est
functions objects
State Model
Design
Observation Model
setpars.survey
generate.sample
sample
4.) Define generating observation model and survey population
Assumed model vs realitygenerate.region generate.density
region density1.) Define survey region & population density
setpars.design
generate.design
design
3.) Generate survey design
point.est int.est
point est interval est
5.) Compute estimates using assumed observation model
setpars.population
generate.population
population
2.) Generate population
functions objects
Assessing sensitivity to assumptionsGenerating observation model defines the true detection/capture probabilities for all animals
Homogeneity equal capture probabilites
Heterogeneitytrap-shyness, unequal exposure
Mark-recapture method
Assumedobservation model
Generating observation model
GenerateEstimate
Assumption violated
Enables sensitivity analysis
Assumed observation model is what we use to carry out the estimation
Function and object names
setpars.survey generate.design plot.design generate.sample summary.sample point.est interval.est point.simetc.
Prefix* (action)
.cr1 capture-recapture
.dp double-platform
.lt line-transect
.no nearest object
.pl plot sampling
.pt point-transect
.rm2 removal
Suffix (method extension)
Example:
setpars.survey.pt Set survey parameters for a point transect
+
1 Estimation functions for .cr extension are .crM0, .crMt, crMb, crMh2 Estimation functions for .rm extensionare .rm, .ce and .cir
* incomplete list
Also summary() and plot() functions for most objects
