S1

Lincoln-Petersen Index S2
M
S 1 S2
N= N 113.40
M
N = Estimated population size
S1 = Initial marked sample size Directions
S2 = Caught sample size
M = Recaptured marked sample size

Assumes:
• The proportion of marked individuals is equal to the
proportion of total population.

S 2 ¿ S1
M¿ to= solve
• Rearranging the equation ¿ for N creates the
Lincoln-Peterson equation.¿ N

• Assumes a closed environment.

Question to Consider:
1) How can the population be estimated if no marked
individuals are recaught?
2) Under what circumstance is M larger than S 1?
27
21
5
 Lincoln-Petersen Index
N = Estimated population size S1
S1 = First sample size, caught and then
S1S 2 S2
marked and returned to the environment
S2 = Size of the second sample N M
M = Number of recaptured individuals in the
second sample
M
N

Population (N)
First Sample (S1), caught, Rearranging the equation allo
(Initially unmarked)
marked, and returned solve for the unknown populat
under the assumption that the
x x x marked to unmarked in both th
x x x population and the second sam
x x x x x S1 = 6
x x x equal.
x x
x
x x
x
x
x
x
Second Sample (S2) S1 S2 ¿
x x =M ¿ ¿
x
x x
x x S2 = 4 N x ¿
x
x x M=1
S1 S2 ¿ 6×4
=M ¿ ¿ N= =24
S1/N = 6/24 = .25
N ¿ M/S = 1/4 = .25 1
2

The ratio of marked to unmarked is equal in both the

experimental population and Sample 2.
dex
25

53
26

50.96

Rearranging the equation allows us to

solve for the unknown population size (N)
under the assumption that the ratio of
marked to unmarked in both the
population and the second sample are
equal.

S1S 2
N
M

6×4
N= =24
1
 S1
Bailey's Modification S2
M
S 1 (S 2 +1)
N B= NB
M +1
NB = Estimated population size
S1 = Initial marked sample size
S2 = Caught sample size
M = Recaptured marked sample size

The Lincoln-Petersen equation tends to overestimate the original population

size and cannot be used if no marked animals are recaught (M=0). Bailey's
modification minimizes the bias to give a more accurate estimate.

Assumes
• Closed environment

• Most effective with sample sizes smaller than 20.

• Individuals are captured one at a time and then released back into the
environment.
• Allows individuals to be counted multiple times.

Questions to consider:
1) What is the estimated population size (N) when no marked individuals are
recaught?
2) Would this be the best method to use when the experimental conditions
require that the individuals in the samples be taken all at once?
 27
21
5

99.00
 Bailey's Modification
NB = Estimated population size S1
S1 = Initial marked sample size S2
S2 = Caught sample size M
M = Recaptured marked sample size

Bailey's Modification

S 1 (S 2 +1)
N B=
M +1
Bailey's modification is a variation of the Lincoln-Petersen estimate. By minimizing the bias, Bailey's modification
for the Lincoln-Peterson equation that tends to overestimate the population size.

Difference in Experimental Design

x
x x
x x x
x x
x x x
x x
x x
x x
x x x x x
x xx
x x x x

The Lincoln-Petersen estimate takes a Bailey's modification samples the population by

group sample so each individual can catching one individual at a time. This allows th
only be caught once per sample. same individual to be caught multiple times with
each sample.
ation
21
21
21
NB 21.00

mizing the bias, Bailey's modification corrects

mate the population size.

Design

x
x
x x
x
x

dification samples the population by

e individual at a time. This allows the
dual to be caught multiple times within
e.
 S1
Modified (Chapman) S2
M

(S 1 + 1 )(S 2 +1) NC
N C= −1
( M +1)
NC = Estimated population size
S1 = Initial marked sample size
S2 = Caught sample size
M = Recaptured marked sample size

This equation is an unbiased estimate of population size.

Assumes
• Closed environment

• Samples are taken all at once so no individual has the possibility of

being counted multiple times.

Questions to Consider:
1) How do the N values from the Modified model compare to the N values
from the Bailey's model with the same values for variables?
 27
21
5

101.67
 Modified
Nc = Estimated population size
S1 = First sample size, caught and then (S 1 + 1 )(S 2 +1)
marked and returned to the environment
N C= −1
( M +1)
S2 = Size of the second sample
M = Number of recaptured individuals in the
second sample

An unbiased population size estimate where the one sample is taken all at once.
 S1 8

S2
18
M 7

Nc 20.38

ken all at once.

x x
x
x x
x
x x x x
x
x x
 S1
Confidence Intervals S2
M
A confidence interval gives the interval that
includes the right answer 95% of the time. It is Lincoln-Petersen
a measure of how close your estimate is to the N 66.67
actual population size. A smaller range means
SN 10.03
that your estimate is closer to the true population
size while a larger range means that your CI 47.01
estimate is not very accurate. Lower

First calculate the standard error:

N 2 ( S 2 -M )
The
√
S confidence
N = interval is calculated by:
( S 2 +1)( M +2)

CI 95%=N±1. 96( S N )
 40
25
15

Lincoln-Petersen Bailey Modified

66.67 65.00 65.63
10.03 9.78 9.87
86.32 45.84 84.16 46.28 84.97
Upper Lower Upper Lower Upper
S1 27
S2 21
M 2

Lincoln-Petersen Bailey's Modified

N 283.50 198.00 204.33
Standard Error 131.73 92.00 94.95
CI 25.31 541.69 17.67 378.33 18.24 390.43
Estimated Population

Comparison of Single Sample Methods

600.00

500.00

400.00

300.00

200.00

100.00

0.00
Lincoln Peterson Bailey's Modified
Method
Modified
 Lincoln Pe Bailey's Modified
CI H 541.69 378.33 390.43
CI L 25.31 17.67 18.24
N 283.50 198.00 204.3333

Method
 Time (t) Ct Rt Ut Mt CtM t
Schnabel 1 20 0 20 0 0
2 20 4 16 20 400
A Schnabel estimate is used when multiple 3 20 3 17 36 720
samples are taken. The first sample is taken, 4 20 2 18 53 1060
marked and returned to the population. In each 5 20 4 16 71 1420
subsequent sample, the marked individuals are 6 20 3 17 87 1740
counted, and the unmarked individuals are 7 20 3 17 104 2080
marked and then returned to the population. 8 20 5 15 121 2420
9 20 4 16 136 2720
Ct = number of sampled individuals at t 10 20 3 17 152 3040
Rt = number of recaptured individuals 31 15600
Ut = number of unmarked individuals (marked
N 503.23
and then returned) N 520.00
t = time

Directions

Σ (C t M t ) when marked > 10% of

N= the population
ΣR t
Mt =ΣU(C
1
+ tUM..U
2 t )t-1 when marked < 10%
N= of the population
ΣR t +1

References
 Time (t) Ct Rt
Schumacher-Eschmeyer 1 20 0
2 20 4
The Schumacher-Eschmeyer model finds the 3 20 3
proportion of total marked individuals at the end of 4 20 2
the capturing period and uses the same variables as
Schnabel. 5 20 4
6 20 3
CM 7 20 3
N= 8 20 5
R 9
10
20
20
4
3
Rearranging the Shnabel equation to solve for R/C
gives:

R 1 Schumacher-E
= ×M
y C= mN* x + b 0.3

Graphing this equation as M = x and R/C = y gives 0.25

1/N as the slope.

R/C
0.2
Taking the inverse of the slope solves for N. f(x) = 0.000724327104406 x +
0.15

0.1

0.05

Directions 0
0 20 40 60 80 1
M

References
 Ut Mt R/C Slope N
20 0 0 0.0218 45.87
16 20 0.2
17 36 0.15
18 53 0.1
16 71 0.2
17 87 0.15
17 104 0.15
15 121 0.25
16 136 0.2
17 152 0.15

Schumacher-Eschmeyer

= 0.000724327104406 x + 0.098502485856335 R/C

Linear
(R/C)

20 40 60 80 100 120 140 160

M
 Bayesian Method
Hypergeo
The one sample models tend to be bias when the number
of recaptured subjects are too low. Bayesian methods are
applied to mark-recapture methods to calculate a more
accurate estimate.

Assumes:
1) A closed population.
2) The probability of capturing a marked subject is equal
to the proportion of marked subjects in the population.
3) Subjects do not lose their marks
Negative B
4) All marks are properly reported.

Mt = total marked animals at the beginning of the tth

sampling interval.
Ct = total number of subjects sampled during t
Rt = number of recaptures in the sample Ct

Calculations
The probability that Ct contains Rt is caluclated using one
Binom
of three methods: hypergeometric, negative binomial, and
binomial.

The posterior distribution is then computed and the

successive distribitions are plotted.
 Reference

Hypergeometric

Hypergeometric

Negative Binomial

Negative Binomial

Binomial

Binomial
 t Ct Mt Ct(Rt) Estimation Method Estimated N
1 34 50 0 Schnabel 8688
2 42 84 1 Schumacher and Eschmeyer 8498
3 43 125 0 Lincoln-Petersen (using t=10) 18832
4 40 168 1 Bailey's (using t=10) 9630
5 32 207 0 Modified (using t=10) 9651.5
6 56 239 1
7 42 294 1 Bayesian Value
8 44 335 4 Mean 10355
9 56 375 3 Median 9700-9750
10 44 428 1 Mode 8650-8750
Population Estimate

Comparison of Methods

20000
18000 Lincoln-Peterson
16000
Bailey
14000
12000 Modified
10000
Schnabel
8000
Schumacher and
6000
Eschmeyer
4000
Bayesian
2000
0
Method

Actual population size = 10,000

 Reference

Example

ln-Peterson

fied

abel

macher and
meyer

sian

= 10,000
 Day 1 2 3 4 5
Total unmarked 4101 1308 1233 1210 1421
Total recaptured 0 318 241 333 463
Total caught 4101 1626 1474 1543 1884
Running total 4101 5409 6642 7852 9273

Recaptures from day: 1 2 3 4 5

1 -- 318 171 153 183
2 -- -- 70 90 99
3 -- -- -- 90 90
4 -- -- -- -- 91
5 -- -- -- -- --
6 -- -- -- -- --
7 -- -- -- -- --
8 -- -- -- -- --

Estimated Population
Population Estimates
Lincoln Schumacher-
Sample Day Schnabel
Peterson Eschmeyer
2 20969 16868 20969 40000
3 34238 20825 33082
4 25279 21316 30777 35000
5 31945 21331 31951 30000
6 26988 20365 30932
7 27559 19372 30817 25000
8 29771 19678 29513 20000
9 29918 16775 30259
Av. 28333 19316 29788 15000

10000

5000

0
2 3 4
 6 7 8 9
Reference
1635 1159 1599 909
700 635 1106 739
2335 1794 2705 1648
10908 12067 13666 14575 Study

6 7 8 9
209 152 269 112
103 71 120 58
103 71 99 67
122 91 117 57
163 98 141 93
-- 152 197 104
-- -- 163 99
-- -- -- 149
Estimated Population

Comparison of Models

40000

35000

30000

25000 Lincoln Peterson

20000 Schnabel
Schumacher-Eschmeyer
15000

10000

5000

0
2 3 4 5 6 7 8 9 Av.
Sample Day
Terms and Conditions:
You may use, reproduce, and distribute this module, consisting of both the software and this associated
documentation, freely for all nonprofit educational purposes. You may also make any modifications to
the module and distribute the modified version. If you do, you must:
• Give the modified version a title distinct from that of the existing document, and from all previous
versions listed in the "History" section.
• In the line immediately below the title, replace the existing text (if any) with the text "© YEAR
NAME", where YEAR is the year of the modification and NAME is your name. If you would prefer
not to copyright your version, then simply leave that line blank.
• Immediately below the new copyright line (even if you left it blank), add or retain the lines:
Original version: Mark-Recapture © 2007 Jessica L. Panks, John R. Jungck
See end of document for full modification history
• Retain this "Terms and Conditions" section unchanged.
• Add to the "History" section an item that includes at least the date, title, author(s), and a description
of the modifications, while retaining all previous entries in that section.

These terms and conditions form a kind of "copyleft," a type of license designed for free materials and
software. Note that because this section is to be retained, all modified versions and derivative
materials must also be made freely available in the same way. This text is based on the GNU Free
Documentation License v1.2, available from the Free Software Foundation at
http://www.gnu.org/copyleft/.

History:
Date: September 17, 2007
Title: Mark-Recapture
Name: Jessica L. Panks, John R. Jungck
Institution: BioQUESTSupport
Acknowledgements: Curriculum Consortium,
for this work wasBeloit College
provided, in part, by the National Science
Foundation Division of Undergraduate Education, the Howard Hughes Medical Institute, and EOT-
PACI.
Modifications: None (original version)