8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
1/18
2. Introduction to distance sampling
Quick introduction to distance sampling and surveydesign in preparation for data collection later today
Cover basic line transects and survey design
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
2/18
Strip transect sampling
0 20 40 60 80 100
0
10
20
30
40
50
0 20 40 60 80 100
0
10
20
30
40
50
0 20 40 60 80 100
0
10
20
30
40
50
0 20 40 60 80 100
0
10
20
30
40
50
Size of study area = 5000 (A)
Total transect length = 50x5 = 250 (L)
Width of half the strip = 1 (w)
Area searched = 2wL= 2x1x250 = 500 (a)
Number of animals counted = 36 (n)
wLa 2
a
nD
ADN D = density
Aa
nN
N = abundance
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
3/18
Strip transect sampling - variance
0 20 40 60 80 100
0
10
20
30
40
50
0 20 40 60 80 100
0
10
20
30
40
50
0 20 40 60 80 100
0
10
20
30
40
50
0 20 40 60 80 100
0
10
20
30
40
50
ni= number of animals seen
onith transect (n=ni)li= length ofi
th transect (L=li)
k= number of transects
var
SE
N
SECV
1var
1
2
k
L
n
l
nlL
n
k
i i
ii
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
4/18
The probability of detecting an animal in the strip is not 1
Collect data that allow that probability to be estimated
Line transect sampling
d
r= radial distance= angle
r
d= perpendicular distance = r sin()
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
5/18
Perpendicular distance
Frequen
cy
w0
Estimating probability of detection
If detection probability = 1in whole strip out to w,
histograms would be here
Animals assumed tohave been missed
Assume probabilityof detection is 1 at
zero distance
ap area under curve
area under rectangle
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
6/18
Line transect abundance estimate
Lw
nD
2
n= number of animals seen
L= length of transect
w= strip half width
Strip transect density (D) estimated as:
pa= average probability of detection
apLw
nD
2
Line transect density estimated as:
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
7/18
Strip width x average probability of detection (wxpa) is
often referred to as the effective strip half width oresw
Effective strip half width
0
5
10
15
20
25
30
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
Perpendicular distance (d)
Frequency
esw
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
8/18
Study area is sampled representatively Equal coverage probability
Animals do not move Randomly
In response to surveyor
All animals are detected on the transect line
i.e. probability of detection at zero perpendicular distance = 1 All measurements are accurate
Species identification
Distances (and angles)
Group size if relevant
Assumptions of line transect sampling
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
9/18
Animals are not distributed randomly in space
Sampling must be random/systematic
To achieve equal coverage probability
Obtaining a representative sample of data
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
10/18
Randomly spaced parallel lines Give equal coverage probability
Survey design
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
11/18
Completely random designs can be logistically moredifficult to implement
And may be more likely to lead to high variance
Survey design
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
12/18
Evenly spaced parallel lines often used in practice With a random starting point
Survey design
But time wasted in transit
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
13/18
Systematic zig-zag lines
With a random starting point
Appropriate for large areas
Survey design
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
14/18
Used for logistical reasons
And when density is known (expected) to vary
Analyse strata separately
Can reduce variance of total estimate
Stratification of survey area
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
15/18
Stratified design with zig-zag lines
Survey lines across density contours
Stratified survey design
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
16/18
Stratified design with zig-zag lines
Survey effort in proportion to expected density reduce variance even more
Stratified survey design
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
17/18
Define the area within which abundance is to be estimated
Search along pre-determined transect lines for animals Effectively searching a strip to either side of the transect
When encounter an animal (group of animals): Identify species
Determine group size
Measure perpendicular distance from animal/group to transect line
Continue searching along transect lines
When finished:
Number of animals encountered is known (n)
Length of total transect is known (L)
Use distance data to estimate average probability of detection (pa)
Line transect sampling in practice
8/10/2019 2_IntroductionDistanceSamplingPSHPL.pdf
18/18
Data organisation and analysis
n= number of animals seen
L= length of transectw = strip half width
pa= average probability of detectionapLw
nD
2
Organise sightings and effort data Analyse in program DISTANCE