USE OF TAGGED- UNTAGGED RATIOS EN
ESTUWAUNG RAEMT PCPULATSONS

Wtesés far 2420 Degree cf M. S.
MECIH‘IEGAN - "231‘? E CGLLEGE
Aerimcé {35:12am Geis

522

This is to certilg that the

thesis entitled

{ I
‘; “Use of t2'25gec‘1-‘231' '13?“de r ties in eet:x.:.ting
J. rabbit eopul'ticns"
l -
presented b1]
’2
" Aelree D. Geis
‘2’ ‘
2.1).! .
i V . .
'.\ v‘ has been accepted towards fulfillment
‘2 ‘ of the requirements for
.1; Lose degree iIFiShEriC‘S C}: 3711(37 ife

.3: 1

flajnr professor

 

r Ihne December 3, 1952
v",
‘41.
0-169
(’ . . ‘, I ‘ r ., 1
~ | t / 1", ' . ‘3‘ ‘ I i
k ./ .- ‘ ‘ ' f

 

USE OF TAGED-UNTAGCED RATIOS IN

ES TIMATING RABBI T POPUIA 'IIONS

By
Aelred Dean “Ggig

ATHESIS

Submitted to the School of Graduate Studies of Michigan
State College of Agriculture and Applied Science
for the degree of

MAS TER OF SCIENCE

Department of Fisheries and Wildlife

1952

'\'_\
~ A
“-r ‘

«.N.

Table of Contents

List of Tables . . . . . . . . . . . . . .
List of Graphs . . . . . . . . . . . . . .
Acknowledgements . . . . . . . . . . . . .
Introduction . . . . . . . . . . . . . . .
Census Methods . . . . . . . . . . . . . .
Field Procedures . . . . . . . . . . . . .

Randomness of Capture. . . . . . . . . . .

Observed compared to expected recapture

Influence of sex, age and previous trap experience.

Nature of heterogeneous trap response
Effect of heterogeneous trap response
estimates. . . . . . . . . . . . .
Correction of biased trap response. .

Area Censused. . . . . . . . . . . . . . .
Summary. . . . . . . . . . . . . . . . . .

LiteratureCited...........o.

1’. Mn
2,3} .‘Trfl!

on population

distribution.

iii

iv

['0

U'l-C'U

ll

13

20
27
29
29
30

Table

Table

Table

Table

Table

Table

Table

Table

Table

l.

2.

3.

h.

5.

7.

8.

9.

List of Tables

Comparison of Binomial and Poisson Distributions

for a Population of 20h Rabbits with a Probability
ofChpmmeof.08. .. .. .. .. .. .. .. .. .
Data Needed for a POpulation Estimate Using a Shot
Sample and to Calculate Probability of Capture. . . .
Comparison of the Observed Distributions of
Recaptures for Two Trap Lines with the EXpected
Binomial Distributions. . . . . . . . . . . . . . . .
Probabilities of Capture of Various Population
Elements on Three Trap Lines. . . . . . . . . . . . .
Comparison to Show the Effect of Combining Data

from the Four Age and Sex Combinations. . . . . . . .
Observed Recapture Distribution Compared to Those
Expected for Rabbits of Various Sexes, Ages and

with Trap Experience. . . . . . . . . . . . . . . . .
EXpected Distribution of Recaptures for a Population
hade Up of Elements with Three Different Probabil-
ities of Capture Compared to That Observed and the
Binomial for a Single Probability of Capture. . . . .
Number of Locations at Which Frequently Handled
Cottontails Were Trapped. . . . . . . . . . . o o o 0
Proportion of the Daily Catch Previously Marked
Compared to the True Fraction of the Population

PTGViOUSly Marked o o o o o o o o o o o o o o o o o o

12

15

18

21

23

iii

Table 9.

Tab le 10 0

Graph I.
Graph I I 0
Graph III.
Graph IV 0
Graph V.

List of Tables Continued

Proportion of the Daily Catch Previously Marked
Compared to the True Fraction of the Population
Previously Marked Trap Line B . . . . . . . . o . .
"Lincoln Index" Population Estimates Based on a
Final Sample Taken in Traps Compared to Those

Obtained When the Second Sample is Shot. . . . . . .

List of Graphs

Observed Compared to Expected Binomial Distribution
of Captures, Trap Line "A". . . . . . . . . . . . .
Observed Compared to Expected Binomial Distribution
of Captures, Trap Line "B". . . . . . . . . . . . .
Observed Compared to Expected Binomial Distribution
of Captures for Several Population Elements . . . .
EXpected Distribution of Captures for a Population
Having Three Probabilities of Capture Compared to
the Observed Distribution and the Binomial Distri-
bution for a Single Mean Probability of Capture . .
Observed Compared to the True Previously Marked

Fraction in Each Day's Catch on Two Trap Lines. . .

2L:

28

IO

16

19

25

iv

Acknowledgements

The writer wishes to express his sincere thanks to Dr. George
°A. Petrides for his direction of the project and his many helpful
suggestions concerning the organization and presentation of this

thesis.

I also want to express my gratitude to Dr. Arthur E. Staebler,

Director of the Kellogg Bird Sanctuary, who provided the writer with

the facilities available at the sanctuary. Dr. Don Hayne's consider—
able help with the statistical aspects of paper was greatly appreciated.
Mr. C. M. McCrary, Director of the Kellogg Station, was very cooperative
in making arrangements for this study to be conducted over the several
units of the Kellogg Station. Er. Walter Lemmien, Forester in charge

of the Kellogg Forest, deserves much thanks for his assistance in set-
ting out trap lines and collecting data from hunters. The financial
assistance provided by a‘W. K. Kellogg Fellowship is also respectfully

acknowledged.

The writer is also indebted to several others who helped in a
variety of ways. They were Dr. Peter Tack, Head of Department of
Fisheries and Wildlife, and Keith Bullock, Ed Caball and Ralph horrill,

employees of the Kellogg Station.

Introduction

As a portion of a study of factors contributing to the high

abundance of cottontail rabbits, Sylvilagus floridanus mearnsii(Allen),

 

on the Kellogg Station of Eichigan State College near Battle Creek,
Michigan, the examination of census methods was undertaken. The cir—
cumstances under which such a study could be made were particularly
favorable during the late fall and winter of 1951 when early severe
winter weather concentrated the rabbit population largely between a
lake and.0pen farm land.‘ This tended to eliminate Ebmplications of
determining the size of the censused area, which would arise if the

study area was part of a larger, more homogeneous plot.

The census method that at first seemed potentially most useful
to wildlife biologists was based on consideration of tagged-untagged
ratios in daily live trapping records. A number of workers have used
live trapping figures to estimate populations by methods which assume
a uniform probability of capture. These include Schnabel (1938) and
Schumacher and Eschmeyer (l9h3) working with fish; Fisher and Ford
(19h?) and Jackson (l9h8) with insects; and Hayne (l9h9) with small

mammals.

Chitty and Kempson (19h?) demonstrated that samples from a partly

marked vole (Microtus agrestis) population were not drawn at random.

 

Young, Neess and Emlen (1952) also found that the house mouse (Egg

musculus) diSplayed heterogeneous trap response. DeLury (1951)

recognized unrepresentative samples caused by marked fish having a
different probability of capture than those that are not marked as

a prdblem.he'was unable to resolve when estimating fish populations

by trapping and marking experiments. then heterogeneity in trap re-
sponse exists, however, population estimates based on trapping results
will be in error due to biased sampling. This was suggested on the
Kellogg Station when population estimates based on trapping did not
agree with others. This indicated that the data which would be expected
to support the assumption of homogeneous trap response should be tested
so that the nature of rabbit trap response could be understood and

accurate population estimates result.

Census Methods
Two ways of determining the tagged—untagged ratio in the population
were used in this study. The first obtained the marked-unmarked ratio
from a sample of shot animals. Allen (1938) used this method to esti—
mate the population by the following formula:

Total number marked Number marked in kill
Total population Total kill

 

 

The other method, as illustrated by Hayne (1949), is similar in
principal except that it is based solely on trapping results. It con-
siders the ratio of marked to unmarked in each day's catch along with
the number previously marked in a cumulative manner to arrive at a
population estimate. This method is based on the assumption that a
uniform probability of capture existed among all members of the

popula tion .

Obtaining a second sample by shooting had several advantages that
would be expected to result in greater accuracy. If biasness exists
in sampling by traps, shooting would probably produce a more representa—
tive sample since it would not involve the same bias. Also, shooting
permits a more complete coverage of the study area. Evidence of the
accuracy obtained by the shooting method was secured when population
estimates from trapping results varied from the total population esti—
mated by shooting to the same extent that the estimate of the population
of hunter—killed rabbits by trapping data only varied from the known
number killed. Therefore, it was concluded that the use of tagged-
untagged ratios in the hunting kill provided a more satisfactory way
of estimating cottontail populations. Many biologists will not be
able to apply this method, however, because they do not have the neces-
sary control over hunting. Hence, it was desirable to further analyze
the use of trapping data alone to estimate rabbit abundance. This was

done by means of a trapping experiment in late 1951.

Field Procedures
Fifty wooden traps (described by Hickie, l9h0) and 27 wire mesh
(size 3, Tomahawk Trap Co., Tomahawk, Wisconsin) were used. To insure
complete coverage the 160 acre study area was divided in half, one
trapped November 3 through 15, the other between November 20 and Decem—
ber 10. An irregular spacing was used because a lake in the center of
the study area with two elongated, curved waterfilled swales leading

from it made the operation of a grid or straight trap lines impractical.

Rabbits were marked by placing numbered tags near the center of
each ear (as described by Haugen, 19h0). There was no evidence of
these tags being lost except that occasionally shot ripped a tag out.
Trap location, age, sex and weight were recorded each time a rabbit
was handled. Closely supervised hunting took place throughout the
entire area between December 15, 1951 and January 10, 1952. The loca-

tion at which each rabbit was shot'was located on a map.

Randomness of Capture

The probability of capture on a trap line represents the average
likelihood that any particular rabbit will be caught on any particular
night. For example, if a probability of capture of .2 exists, the
chances are 2 in 10 that a certain rabbit will be captured on any night,
or in 10 nights the rabbit will be eXpected to be captured twice. It
is computed by dividing the number of times captures are made on a trap
line by the number of nights the line'was operated times the population
present. For example, if 200 captures were made on a line run for 10
nights with 100 rabbits in the vicinity, the probability of capture
(p) would be calculated as follows:

p _ 200 _ .2

10 x 100

If there is a uniform probability of capture among all members
of a population or, stated another way, if each capture represents
a random sample from the population, then the distribution of the
number of times different members of the population are captured
should agree with a poisson or binomial distribution. Snedecor (l9h6)

and Simpson and Rowe (1939) do not give concrete rules as to when the

poisson or binomial distribution should be used. The essential
difference in the two is that the poisson distribution is used where
a very small probability of the occurance taking place exists. that
constitutes a low probability, however, is not defined. To determine
if a significant difference exists in distributions calculated by the
two methods, a theoretical distribution was calculated by both methods
for a probability of a magnitude commonly encountered in the data

(p I .08). This comparison showed that the two methods gave very
similar results (Table 1). Throughout this study the suggestion of
Ricker (1937) has been followed that the binomial distribution be
calculated when a probability of .05 or greater exists using the

poisson for smaller probabilities.

Observed compared to eXpected recapture distributions. To determine

 

if rabbits were captured in a random.manner on the study trap lines
the observed distribution of the number of times individuals were
captured was compared to the expected binomial distribution. The
observed frequency of capture of marked rabbits was obtained from
their trapping records. But, in order to determine how many rabbits
were not captured it was necessary to estimate the total population
from the tagged-untagged ratio in the hunting kill (Table 2). Evi-
dence that estimates based on the kill are accurate has been already
given. From the comparison of observed with eXpected values (Table 3
and Graphs l and 2) it is apparent that more rabbits were captured
in the zero and in the higher categories (3 and up) than would be
the case if a uniform probability of capture existed. Chi—square

tests indicated that these differences Were highly significant and

TABLE 1

COMPARISON OF BINOMIAL.AND POISSON
DISTRIBUTIONS FOR.A POPULATION OF 20h RABBITS
WITH A PROBABILITY OF CAPTURE OF .08

 

 

 

 

 

a=========;
Times EXpected Values
Captured
Binomial Poisson
0 68.95 71.81
1 78.07 7h.98
2 110.73 39.11
3 12.95 13.62
h 2.82 3.56
S .hh .7h
6 .05 .13
7 .OO .02
8 '.OO- .00

 

20h.01 20h.01

 

TABLE2

DATA NEEDED FOR A POPULATION ESTIMATE
USING A SHOT SAMPLE AND TO CALCULATE
PROBABILITY OF CAPTURE

 

 

 

 

|_Trap Line

[ A B
Number marked 79 89
Total number shot 71 65
Marked rabbits shot 28 23
Estimated population 200 251
Total captures 213 160

Probability of capture .082 .053

 

COLLPARISON OF Tn'E OBSERVED DISTRIBUTIONS

TABLE 3

OF RECAPTURES FOR TWO TRAP LINES WITH
THE EXPECTED BINOLxlAL DISTRIBUTIONS

 

 

 

 

 

 

 

Number Trap Line A Trap Line B

Capifires Observed Expected Observed Expected
Number Number Number Number

0 121 67.60 163 135.5h

1 3S 76.5h 60 85.5h

2 12 39.93 12 2h.80

3 6 12.70 6 h.3h

u 11 2.76 S .63

S 8 .113 l .113

6 3 .Oh 2 .03

7 0 .00 2 ..01

8 3 .00 1 .00

9 0 .00 o .00

10 1 .00 0 .00
200 200.00 251 251.00

 

GRA PH I
Observed compared to eXpected binomial distribution of

captures, trap line "A".

NUHIER OF RABBITS

GRAPH I

 

'20 - ———oas£nv:o
———-:xpecno

IOO »

30 ~

 

40* ‘

 

20P

 

 

 

NUMBER OF-TINES OAPTWED

GRAPH II
Observed compared to expected binomial distribution of

captures, trap line "B".

RABBITS

NUMBER OF

I60

 

I40 -

I20 -

1

I00

20..

 

GRAPH II

 

 

OBSERVED

‘ “’“"EXPEOTED

 

 

 

 

NUMBER 08. TIMES CAPTURED

lO

11

that the distribution of captures were not at random. Hence, the
assumption basic to current census methods using only trapping data,
that a uniform probability of capture exists, has been shown not to

apply to Kellogg Station rabbits.

Influence of sex, age and previous trap experience. In further analysis

 

to understand and possibly correct for this variance, the data for

three trap lines were broken into 16 identifiable elements as to sex,

age and previous trap experience. The probability of capture was cal-
culated for each to determine if the different population elements
consistently had different probabilities of capture of the same direction
and magnitude. Table h shows that differences did exist between cate-
gories but they were not consistent except in regards to trap eXperience.
Once a rabbit was captured it was more likely to be recaptured than a

rabbit which had not been taken.

However, differences in other categories did exist and it is
possible that the distribution of recaptures in a sample made up by
pooling trapping data from several population elements with different
probabilities of capture will not correSpond to a binomial distribution
even though the sampling within each population element has been at
random. In order to determine if the discrepency between eXpected
and observed values was due to the various combinations of ages and
sexes having different probabilities of capture, the following analysis
was made. For each age and sex combination the theoretical expected
fraction of the sample for each capture category was calculated (Table

5). These were then combined in the same ratio as the numbers of

TABLE h

PROBABILITIES OF CAPTURE OF VARIOUS
POPULATION ELELENTS ON THREE TRAP LINES

 

 

 

A B C

Population Element Nov. Dec. Karch
Combined .082 .053 .0h6
A11 males .08h .037 .Oh8
All females .079 .062 .Oh5
All adults .111 .050 *
All juveniles .077 .052
Adult males .08h .Ohh
Juvenile males .O8h .037
Adult females .131 .052
Juvenile females .071 .O6h
All after first capture .lhl .072 .070
hales after first capture .177 .067 .076
Females after first capture .116 .075 .065
Adults after first capture .131 .017 *
Juveniles after first capture .lhl .091
Captured first on a previous line .133 .039
Not captured on a previous line .062 .060

 

* In March all are considered to be adults.

13

individuals in the various age and sex combinations. These weighted
mean values (Table 5, line 5) were then compared with those obtained
by calculating the expected binomially distributed fractions directly
from the probability of capture for the entire population where the
different ages and sexes were not considered separately (Table 5,
line 6). There was not a large enough difference between the two
series of values to account for the large differences previously
noted. This indicates that the discrepancies were not due to age and

sex differences.

Further evidence that the discrepancy is not due to a random
distribution of captures Within each age and sex combination having
a different probability of capture is shown in Table 6. There the
observed distribution of captures within each age and sex group and
of rabbits after they have been captured once is compared with what
would be expected if the distribution was at random. These compari—
sons also are displayed in Graph 3. Chi-square tests revealed that
significant differences occurred in each instance. AS'With the comp
bined data, there were too many individuals in the no- and many-capture
categories. This rather conclusively demonstrates that combining
data from rabbits' different ages, sexes and trap experience was not
the cause of the variance between the Observed and expected values.
Trap addiction or avoidance apparently is an attribute of the individ-

ual rabbit which results in non-random selection of animals by traps.

Nature of heterogeneous trap response. The above discussion demonstrated

 

the existence of heterogeneous trap response, but it gave no indication

 

 

 

 

000. N00. mHOo JOC- JON. 0mm. ammo N®Oo Goflfinfifimflv
HdfifiOGHm
OOO. MOO. wHOo ONO. NON. Q. m. Hmm. NmO. hmumm um um
Basso
So. So. 80. as. m5. Sm. an. no. an m £283
000. MOO. BO. ONOo OHN. Ham. Ode “.30. mm .3 0.330de
m8. 0H0. 000. mm...... 8N. mm. H9”. H2. m G Pg“;
000. m8. 0H0. ONO. OHN. Ammo CNN. .50. m e” 9436...“.

s m a m N H o d
_ HMHMWaMw ooh—peg om¢
homemade despise some 5” pepooawo sowpodaom 65 Mo 5.3.0me IHHnmnmEnw .3852 one New

 

 

 

mZOHB¢sz§oo 3m 93. ”mom finch m5. acxh
add OZHZHmEOO mo sonogram Neg .115me 2. zomHmdngoo

m mama;

15

 

oo. o oo. o 8. H 8. 0 0H
8. H 8. o oo. o oo. o a
oo. o 00. H oo. H oo. H s
so. n oo. o oo. o oo. o a
on. o oo. o co. m «o. o o
so.H m 0H. m mm. m ow.. H m
me.m a mm. : mm.H a mm. m a
mm.mH HH sm.s N Hm.m a .oo.m H m
ao.mm o mm.eH o m:.sH. a mm.s o N
om.mm NH m:.mm 3 «0.3 m omé m H
mm.0H mm m~.mm om sm.sm mm ~:.m HH 0

 

panacea Eggnog @3005 “US$63.04 oopoonwm oofimmob _ c3005 oozompo nohfiaoo
mo

 

 

0.3.38 25 98:21— uoawaom 92...“?pr % modem 3325i. _ 3.264 - $852

 

 

mUZMHmemxm damH ers.aza mesa .mmamm mssHmas so
maHsmae mos msHommxm macaw oh amsamsoo ZQAassseHmHm expadasas mm>eammo

0 m4 m3.

GRAPH III
Observed compared to expected binomial distribution of

captures for several population elements.

 

GRAPH Ill

JUVENILE MALES

 

 

O E 4 B B IO
TINEB 'OAPTUREO

AOULTB

 

JUVENILE FENALEB

 

 

 

ALL RABBITB
AFTER ONE CAPTURE

 

 

 

17

as to the extent tO‘WhiCh the probabilities of capture varied or the
relative numbers of individuals with different prObabilities of cap-
ture. An examination of the distribution of recaptures on a trap

line suggested that they might be divided so as to fall into three
groups with different probabilities of capture. One hundred and
sixty-eight rabbits were caught 0-2 times, 25 were captured 3-5 times,
and 7 were taken 6-10 times. The probabilities of capture of these
groups were .027, .31h and .571, respectively. however, for the above
classification to be valid the expected binomial distribution Of cap-
tures for each of these trap vulnerability categories when added
together must closely approximate that Observed from actual trapping
records. TO test this the expected number of individuals in each
capture category i.e., caught 0 times, 1 times, 2 times, etc., was
calculated for each trap vulnerability classification (Table 7, col—
umns l, 2 and 3). Then the eXpected values in each capture category
for each trap vulnerability classification were added to Obtain the
number of individuals in each capture category for the entire popu-
lation made up of representatives from.three elements with different
probabilities of capture (Table 7, column h). There was no signifi-
cant difference between this distribution and the one based on the
actual trapping records. Also, as before there was a highly signifi—
cant difference between this distribution and that Obtained by assuming
that the population was made up of individuals having an equal probabil-
ity of capture. The relationship between these three distributions

is shown in Graph h. This analysis indicates that the reaction of

the rabbit population to traps can be eXplained by recognizing that

 

 

18

 

 

0.08 08 0.00N 0; 0.mN -i 0.03
H.. H. NH
N... N. HH
H 0. 0. 0H
0 N.H H.H H. a
m H.N m.H 0. 0
o 0N m.H 1...H N
m 0.: H.H 0N 0
s. 0 am 0. 0.: m
0.m HH N0 N. Rm .N. a
0.NH 0 0.0 H. H.m 0. m
was NH m.0H 0.m m; N
N.t. mm 0.3 H.H m4: H
0.3 HNH 1QHH N. N.0HH 0
HN.u0 mind NN0.u0
are he News. is. $02 2e... 2.:
ooehoapo mo sum mofifle

oonapmmo on Op oobhombo
3392 00 203025.03 0300.00.

 

 

 

 

mxbemdo mo HEHqudmomm muozam < mom A<Hzozam may 22¢ Qm>mmmmo
93.1? 8. gfiaoo mimmewo mo mm; HAHmdmoxm .w zfiszmHm mama. EH3
mezmamqm m6 m: and: ZOHB<ADAOQ «.mom mambam<omm mo ZOHBDmHmHmHQ QHHommxm

u .593.

,-

Unit;

“(1

R IV

EXpected distribution of captures for a population
having three probabilities of capture compared to the
observed distribution and the binomial distribution for

a single mean probability of capture.

 

 

aunts our means

120 '

I00

0
O

O
O

 

l9

GRAPH m

BINONIAL DISTRIBUTION WITH THREE
PROBABILITI‘IB OF CAPTURE

..............

_..______ BINOMIAL DIBTRIBUTIOR WITH OR!
PROBABILITY OF CAPTURE

 

O BBBRVEB =0IBTRIBUTION

 

 

3 4 B B 7 B 9 IO
NUMBER OF TIMEB CAPTURED

20

the pOpulation is made up of individuals with varying probabilities
of capture. Some have a high probability and a relatively large
number have a low probability. Other individuals have an intermedi-

ate position.

The question may be raised as to whether those animals caught
many times were trap-addicted i.e., had something in their nature
which caused them to readily enter traps, or whether they merely had
a small range near a trap and consequently were captured frequently.
Table 8 sheds light on this question. It shows that the frequently
captured rabbits had been caught in a number of different locations.
This suggests that these rabbits had an inclination to enter traps.
Possibly they had a particular liking for the corn used as bait, or
perhaps merely a curiosity about traps, and consequently entered them
whenever they were encountered. Other evidence of the addiction to-
wards traps was shown in tracks in the snow of rabbits walking around
Sprung traps after the trapping period had been completed, apparently
trying to enter them. It is not believed rabbits sought traps for
cover because they apparently entered open wire mesh traps as readily

as solid wooden ones.

Effect of heterogeneous trap response on population estimates. The

 

above consideration indicates that any census method based on trapping
records which assume a uniform probability of capture will lead to
inaccurate rabbit population estimates. Also, these estimates will

always be lower than the actual population because:

TABLEB

NUMBER OF LOCATIONS AT WHICH FREQUENTLY
HANDHELD COTTON TAILS 'IIERE lMPfED

 

 

 

Number of
Rabbit limes Captured Trap locations
A 10 S
B 8 7
C 8 6
D 6 5
E 6 3
F 5 h

 

21

22

l. The members of the population which have a high
probability of capture tend to cause a high prev—
iouslyhcaptured fraction in each catch and hence
depresses the population estimate.

2. The largest part of the population is not captured
during the trapping period and consequently is not

given due consideration.

Trapping records are commonly used in three ways to obtain
population estimates. One way is to consider the marked—unmarked
ratios during parts of a continuous sampling period. There are
several modifications of this method, but all assume random sampl-
ing. If this exists then the previously marked fraction in each
day's catch should represent the true marked fraction in the popula-
tion. The extent to which heterogeneity of trap response distorts
the previously marked fraction in each day's catch from the correct
value is shown in Table 9 and Graph h. The end result of this dis-
crepency is shown when the estimated population from trapping data by
the method proposed by Hayne (l9h9) is compared with that Obtained
fromLthe marked-unmarked ratio in the hunting kill. The latter method
avoids the bias in sampling present in the first method based solely
on trapping. Trap line "A" had a population estimate from trapping
data which was h0% of the much more accurate estimate from shot animals.
Trap line "B" had an estimate which was h7% of the number arrived at
from the tagged-untagged ratio in the hunting kill. Probably a more

accurate way of determining the extent to which the population estimate

 

23

 

mam. mom . E NN 0N N H ._
mom. me. ms mm mH : :H =
ozm. wmw. we mH 4H m MH =
OHm. mew. mo mm Hm 0 «H =
00N. mHs. Nm 0N .sH 0H HH ..
QHN. 0mm. m: mm NH OH OH =
msH. New. mm (Hm 4H s m =
qu. ems. 0N HH m w m =
0NH. fie. :N a a m a ..
owo. 00m. wH mH m «H 0 z
0m0. mmm. 0H m H N m ..
040. mmm. a m H N s ..
000. 000. o m o m m .>oz
voHpnmm vapnmm coHpcmm ham maHe pmaHm
hHm30H>mHm hHm50H>oam hHm=OH>mhm you nopmo how
COHpmHSQOm mo nopmo mo nomadz Hmpoe Hmpoe mondpdwoox UmHocmm 09mm
GOHpHOQOhm GOthomoam

 

 

«uEfiHmfifi

Dflmx<£ MAmQOH>£xm ZQHHHHDLOM my“ ho ZQHHudfim MD£B
mmw OH ammdlaoo szx<H MHmDOH>mmm rgw<o NAH<Q MEH mo ZOfibeAOmm

m mnmde

2h

 

 

 

 

Omm. mom. we HH OH H OH =
mam. me. Om HH m m m ..
mNm. mam. H0 NH a m 0 ..
Sm. mg. 00 0 a H a ..
mom. mNs. 0a a m a 0 ..
3N. flN. m0 fl m HH m ._
mmm. men. _ 00 m m. m : .oom
MON. $2. Hm mH N. m mm. ..
msH. mmm. mg NH s 0 mm =
mMH. 3:. mm H.H H. OH mm ..
mac. $0. 4N NH H HH HN ..
GOO. 000. 0 4m 0 gm Om .boz
032mm emHesmm 032$ .30
hHm50H>mhm hHmsOHbmhm hHmdofiEHm .3,“ £088 ,
coflmHsdom mo anO we pmgz H309 H309 mmhspdmoom Roz 3.0m
:OHp90doam GOH¢h0d0Hm
m mZHH mmdy

podepcoo m wqmgw

 

r ..I \(llt'l‘

 

GRAPH V
Observed compared to the true previously marked

fraction in each day's catch on two trap lines.

PREVIOUSLY MARIED FRACTION

25

GRAPH Y
I.O V LINE “A“
. . - B
I g:
.3 . ° . OBSERVED

.4 .
. .. z

 

 

 

 

TRUE VALUES
.2 .- O
o A n 1 4_ A 4 L n n .1
0 go .90 .0 BO IOO
I.O F TRAP LINE '9'
I» '
,8 ..
. {OBSERVED
.6 ’ o '
b O
A .. ° '
O
2? .' RUE VALUES
0 l l 1 4 A 1 l l J J
20 40 so so IOO

TOTAL NUMBER MARKEO_

26

was distorted due to heterogeneous trap response is shown by estimating
the population of shot rabbits from their trapping record. In other
words, a popuLation estimate for the number killed was made from the
trapping records of a known number of shot cottontails. These turned
out to be h3% and h2% of the number shot in the vicinity of trap lines
"A" and "B", respectively. This not only shows the inaccuracy of the
estimates based solely on trapping data but also indicates that the
estimates of the total population based on the tagged-untagged ratio
in the hunting kill were prdbably quite accurate. Data from the Kel-
logg Forest, although not collected with the same precision as that
at the Kellogg Bird Sanctuary and Farm, indicate that estimates based

on trapping results also ran about h0% of the number present.

Haugen (l9h0) estimated rabbit populations by trapping until
previously marked animals predominated in each day's catch. He then
considered the number that had been marked as the total resident popu-
lation. Later captures of unmarked rabbits were described as being
transients which were not part of the resident population. 'hhen
Haugen's method was applied to data collected in this study, the
population estimates were only 28.5% and 39.5% of the estimates ob-

tained from the tagged-untagged ratio in the hunting kill.

Green and Evans (l9h0) with snowshoe hares and Southern (l9h0)
with the European wild rabbit used the tagged-untagged ratio in a sec-
ond trapping period only to estimate the number present. For use with
the cottontail this practice yields estimates that are far too low.

This is illustrated by estimating a population using the average

27

marked fraction for the last two days of trapping to represent the
fraction of the population that had been previously marked. This
lestimate can then be compared with that obtained by trapping, then
shooting (Table 10). Once again the estimated number was only about
h0% of the actual number present. It is apparent that population
estimates which depend upon a second large sample by trapping are just
as inaccurate as those in which each day's catch is given consideration

in making the estimate.

Correction of biased trap response. Because population estimates by

 

the method described by Hayne (l9h9) consistently ran about .hO of

the number present it seems justifiable to use the reciprocal of .hO
or 2.5 as a correction factor for population estimates based solely
on trapping results. The spring 1951 breeding adult population of the
Kellogg Bird Sanctuary and Farm based on trapping results was 75, an
obviously low value. If that number is corrected by multiplying by
2.5 the estimate would be 188, a much more reasonable number judging

from the previous spring's population and field observations.

Although the above correction factor appears to hold at the
Kellogg Station thus far, it should be tested more widely. The in-
fluence of the time of year, various trap spacings and different
population densities should also be evaluated. How constant the
correction factor remains under different conditions depends on how
constantly the probabilities of capture'within the population vary

to the same relative extent.

28

TABLE 10

"LINCOLN INDEX" POPULATION ESTIMATES BASED ON A FINAL
SAMPLE TAKEN IN TRAPS COMPARED TO THOSE OBTAINED
WHEN THE SECOND SAMPLE IS SHOT

 

 

 

Trap Line A Trap Line B
Number marked up to last two days 73. 86.
Average marked fraction last two days .863 .863

Estimate population
final sample by trapping 8h.6 99.7
N " " shooting 200. 251.

 

29

Area Censused

In order to determine population densities it is necessary to
know the area over'which animals are being censused. This problem
was simplified in present study because the population was largely
concentrated in winter cover which lay between open farm land and a
lake; and the locations of marked shot rabbits known. By marking the
locations on a map where tagged rabbits were shot, it was possible to
see at a glance the area over which marked rabbits ranged. Unfortunately
other workers may not enjoy these benefits and consequently determining
the area censused may be a very complex problem. Several mammalogists
(Dice, 1938; Stickle, l9h6; MacLulich, 19Sl)'working with small rodents
have offered solutions to this problem based on capture locations.
Unfortunately, the assumptions basic to these methods are questionable
when working with rabbits. The rabbit research project at the Kellogg
Station has not yet developed a method of estimating the census area
based solely on trapping records; however, the subject is being investi-
gated.

Summary

Rabbits were not live-trapped in a random manner. This was not
due to age or sex. The inclination to enter or avoid traps apparently
was a quality of the individual rabbit. The nature and extent of

variation in trap response was demonstrated.

The error in several current census methods caused by heterogeneous
trap response was shown and a method of correction for it suggested.
Evidence was presented which indicated that population estimates based

on the tagged-untagged ratio in the hunting kill were accurate.

 

 

 

 

30

Literature Cited

Allen, Durward L. 1938. Ecological studies on the vertebrate fauna
of a 500 acre farm in Kalamazoo County, Michigan. Ecol. Monog.
8: 3h7-h36.
Chitty, Dennis and D. A. Kempson. l9h9. Prebaiting small mammals
and a new design of live traps. Ecology 30: 537—5h2.
DeLury, D. B. 1951. On the planning of experiments for the estimation
of fish populations. J. Fish Res. Ed. Can. 8: pp. 281-307.
Dice, Lee R. 1938. Some census methods for mammals. Jour. Wildl.
Mgt. 2: 119-130.
Fisher, R. A. and E. B. Ford. l9h7. The spread of a gene in natural

conditions in a colony of the moth Panaxia dominula. Heredity

 

l: 1&3-l7h.

Green, R. G., and C. A. Evans. l9h0. Studies on a population cycle
of snowshoe hares on the Lake Alexander area, I. Gross annual
censuses, 1932-1939. Jour. Wildl. Mgt. h: 220-238.

Haugen, Arnold O. l9h2. Life history studies of the cottontail
rabbit in southwestern Michigan. Amer. Midland Nat. 28: 20h-2hh.

l9h3. Management studies of the cottontail rabbit

 

in southwestern Michigan. Jour. Wildl. hgt. 7: 102—119.

Hayne, Don W. l9h9. Methods for estimating pepulation from trapping
records. Jour. Mamm. 30: 399-hll.

Hickie, Paul. l9h0. Cottontails in Michigan. Game Div., Michigan

Dept. Cons. pp. 1-112.

31

Jackson, C. H. N. 19h8. The analysis of a tsetseqfly population, III.
Ann. Eugenics 1h: 91—108.

MacLulich, D. A. 1951. A new technique of animal census, with examples.
Jour. Eammm 32: 318-328.

Ricker, W. E. 1937. The concept of confidence or fiducial limits
applied to the poisson frequency distribution. Journ. Amer. Stat.
A880. 32: 3h9-356.

Schnabel, Z. E. 1938. The estimates of the total fish population of
a lake. Amer. Math. Monthly us: 3h8-352.

Schumacher, F. X. and R. W. Eschmeyer. l9h3. The estimate of fish
population in lakes or ponds. Journ. Tenn. Acad. Sci. 18: 228-2h9.

Simpson, G. G. and A. Rowe. 1939. Quantitative zoology. thraw-Hill
Book Co., New York and London. xvii + hlh pp.

Snedecor, George W. l9h6. Statistical methods applied to experiments
in agriculture and biology. Iowa State College Press, Ames, Iowa.
xvi +-h85 pp. ‘—

Southern, H. N. 19hO. The ecology and population dynamics of the wild

rabbit (Oryctolagus cuniculus). Ann. Applied Bio. 27: 509-526.

 

Stickle, Lucille. l9h6. Experimental analysis of methods for measuring
small mammal populations. Journ. Wildl. Ngt. 10: 150—159. ‘

Young, Howard, John Neess and John T. Emlen, Jr. 1952. Heterogeneity
of trap response in a population of house mice. Jour. Wildl. mgt.

l6: 169‘1800

 

.. . . . . .. . .... ‘llr. H. .. 1..
. . . . .«L 3.5.. ,. ., V
. Amati. .

an,» _. .. .. emu"-urk..........~wn- .. _ ..
a V. ., .- . an adieu}. d}: are... .4...
. x . .. d . . .. I IN ... . . v . ..... L
a .. Hwaflsmwfi. .. L3}. as» 3...“ fit _
. s I _ II. . 3.0.x. at! _ . ...- r. I...IMI.B; .
.3\_..... h. .b‘xu. . ‘2 r .. .. . U. \ m “Hf; .‘_..d)vl.\‘fl . fl . . .

WM

4
r ,
b

my

use can

M I ' . c u l.‘.
B ...- , -.~\. 0 .r v
r . r. xii“? ..\u... . _
. 1.1.! I. .. my...» a... .1 .. .
c H u (v ., 30" .1.‘ .., .1 ..I y . .. .
'Jl I u Bu..‘ I ,. x _ . BI. .
I 1" .1. a . 74 .9: H ... . u .. II. iv an .
.. p. a. .4 .. . .. . .. . . , .. H.. .
.. , ,, . I . , . . .. . q . .,
R . L. . .
. . v o c
u I
V, B B B J '
u
p
.\ o
r
I a I . _
Q
l ‘ O
l u
o
u I I
u
' .
. J u .
I \ a I I
l n _
B V ' I l
c . I r
. _
o n o 0
R I O
. a.
.l I A l . A O _
0
.
a I
I ‘

 

MICHIGQN STQTE UNIV.

H | II I ll"

LIBRQRIES

Ill!

 

31293100602600

.‘Ii