ESTIMATING NUMBERS

Basic to the study of animal populations is the estimation of their

numbers, no small task in wild populations. During the past several

decades, much work has gone into the development of techniques

and statistical methods to arrive at some estimates of animal

populations. Basically the methods of estimating the numbers of

animals can be put into three categories: true census, a count of

all individuals on a given area; sampling estimate, derived from

counts on sample plots; and indices, the use of different types of

counts, such as roadside counts, animal signs, and call counts, to

determine trends of populations from year to year or from area

to area.

True Census. A true census is a direct count of all individuals in a

given area. It is difficult to do for most wild populations, but

there are situations where a total count can be made.

Many territorial species are easily seen and heard and can be located in

their specific area. Such a census is regularly used for birds. The spot-map

method is probably the best approach. A sample plot of at least 10 hectares

is marked out in a grid with numbered stakes or tree tags placed at intervals

of 50 m.

1
Five or more daily counts are made throughout the breeding

season. Each time a bird is observed, it is marked on a map of the

plot. At the end of the census period all the spots at which a

species is observed are placed on one map. The spots should fall

into groups, with each group indicating the presence of a breeding

pair.

The groups for each species can then be counted in order to

arrive at the total population for the given area. Results are

usually expressed as animals per hectare.

Direct counts can be made in areas of concentration. Deer in open

country, herds of elk and caribou, waterfowl on wintering

grounds, rookeries, roosts, breeding colonies of birds and

mammals permit direct counting usually either from the air or

from aerial photographs.

Sampling Estimates. The sample estimate of population size

involves two basic assumptions:

(1) the mortality and recruitment during the period the data

are being taken are negligible or can be accounted for;

and

2
 (2) (2) all members have an equal probability of being

counted—they are not trap-shy or trap-addicted, they

are distributed randomly through the population if

marked and released, and they do not group by age, sex,

or some other characteristic.

Sampling also involves one major general consideration. The

method employed must be adapted to the particular species, time,

place, and purpose.

Relatively immobile forms, such as barnacles, mollusks, and cicada

emergence holes, can be estimated by the quadrat method,

similar to that employed for plants. The data can be analyzed for

presence, frequency, and so on, or the results can be converted

to a density per hectare. The size and shape of the quadrat will

depend upon the density of the population, the diversity of the

habitat, and the nature of the organism. A few preliminary sur-

veys are made before settling on a quadrat size.

Foliage arthropods may be sampled by a number of strokes with a

standard sweep net over a 10-in2 area. The number of strokes

needed to secure the sample must be predetermined. It will vary

with the type of vegetation.

3
Estimates of zooplankton, obtained by pulling a plankton net

through a given distance of water at several depths, can be made

by filtering a known volume of sample through a funnel using a

filter pump. The filter paper should be marked off in equal

squares. With the aid of a hand lens or a binocular microscope,

the organisms in each square can be counted. The numbers then

can be related back to the total volume of water sampled.

CAPTURE-RECAPTURE METHODS

Introduction

None of the numerous techniques available for estimating the

size of animal populations is foolproof, and none can apply equally

well to all populations. This section presents a popular method

useful for estimating the population size of a single species of

highly mobile animal, such as most small, medium and large

vertebrates.

It is called the capture-recapture, or mark-recapture,

technique. In honor of some early contributors to its

4
development, fishery biologists refer to the basic procedure as

Petersen’s method. ornithologists and mammalogists call it

Lincoln’s method, and diplomatic ecologists refer to it as the

Lincoln-Petersen method.

Historical Background

One way to estimate the size of a population is to capture and

mark individuals from the population, release them, and then

resample to see what fraction of individuals carry marks.

John Graunt first used this simple principle to estimate the

human population of London in 1662. The first ecological use of

mark-and-recapture was carried out by Danish fisheries biologist

C. G. J. Petersen in 1896 (Ricker, 1975).

Tagging of fish was first used to study movements and migration

of individuals, but Petersen realized that tagging could also be

used to estimate population size and to measure mortality rates.

Fisheries biologists were well advanced over others in applying

these methods.

5
Lincoln (1930) used mark-recapture to estimate the abundance of

ducks from band returns, and Jackson (1933) was the first

entomologist to apply mark-recapture methods to insect

populations.

The strength of mark-and-recapture techniques is that they can

be used to provide information on birth, death, and movement

rates in addition to information on absolute abundance.

The weakness of these techniques is that they require

considerable time and effort to get the required data and, to be

accurate, they require a set of very restrictive assumptions

about the properties of the population being studied.

Mark and recapture techniques may be used for open or closed

populations. A closed population is one that does not change in

size during the study period, that is, one which the effects of

births, deaths, and movements are negligible.

An open population is the more usual case, a population that

changes in size and composition from births, deaths, and

movements. Different methods must be applied to open and

closed populations.

6
 1. Closed populations, single marking – Petersen method

2. Closed populations, multiple marking – Schnabel method

3. Open populations, multiple census – Jolly-Seber method

PETERSEN METHOD

The Petersen method is the simplest mark and recapture method

because it is based on a single episode of marking animals and a

second episode or recapturing individuals.

The basic procedure is to mark a number of individuals over a

short time, release them, and then recapture individuals to check

for marks.

The second sample must be a random sample for this method to

be valid; that is, marked and unmarked individuals must have the

same chance of being captured in the second sample. The data

obtained are

M = Number of individuals marked in the first sample

C = Total number of individuals captured in the second sample
R = Number of individuals in second sample that are marked

From these variables, we need to obtain an estimate of

N = Size of population at time of marking

7
By a proportionality argument, we obtain

N=C
M R

Or transposing

N = CM (1)
R

Where N = estimate of population size at time or marking

M = Number of individuals marked in first sample
C = Total number of individuals captured in second
sample
R = Number of individuals in second sample that are
marked

Example; Estimation of the population size of Rattus norvegicus in

1 ha plot.

M = 200 Rattus norvegicus

C = 250
R = 50

From Equation 1

N = CM
R
= 250 x 200
50
= 1000

8
This formula is the “Petersen estimate” of population size and has

been widely used because it is intuitively clear. Unfortunately,

formula (1) produces a biased estimator of population size,

tending to overestimate the actual population. This bias can be

large for small samples, and several formulas have been suggested

to reduce this bias.

Chapman (1951) and Seber (1982) recommends the estimator

N = (M + 1)(C + 1) - 1 (2)
R+1

Which is unbiased if (M + C) > N and nearly unbiased if there are

at least seven recaptures of marked animals (R > 7). This formula

assumes sampling without replacement in the second sample, so

any individual can be counted only once.

In some ecological situations, the second sample of a Petersen

series is taken with replacement so that a given individual can be

counted more than once. For example, animals may be merely

observed at the second sampling and not captured.

For these cases the size of the second sample (C) can be even

larger than the total population size (N) because individuals might

be sighted several times. In this situation we must assume that

9
the chances of sighting a marked animals are on average equal to

the chances of sighting an unmarked animal. The appropriate

estimator, from Bailey (1952), is

N = M(C + 1) (3)
R+1

Which differs only very slightly from equation (2) and is nearly
unbiased when the number of recaptures (R) is 7 or more.

Equation (3) is appropriate if the second sample of animals is

obtained by collecting the animals one at a time, returning each to

the population before taking another (which is analogous to

removing a ball from the pot and returning it before taking

another). This means that an animal (or ball) can be counted more

than once as a member of the second sample (contributing more

than once to the determination of R).

Equation (2) is a more typical situation is where the animals of

the second sample are captured all at once, so individual can only

be counted once as a member of that sample.

N = (M + 1)(C + 1) - 1 (2)
R+1

10
N = (200 + 1)(250+1) - 1 = 988
50 + 1
and

N = M(C + 1) (3)
R+1

N = 200 (250 +1) = 984

50 + 1

As with all population estimates made from samples, there is an

uncertainty caused by the error associated with examining a

sample rather than the entire population. A measure of this error

that expresses the uncertainty of a capture-recapture population

estimate is the standard error (SE).

For Equation (2), this computed as

SE = (M + 1) (C + 1) (M - R) (C – R) (4)
(R + 1)2 (R + 2)

SE = (200 + 1) (250 + 1) (200 – 50) (250 – 50) = 105.8

(50 + 1)2 (50 + 2)

11
For Equation (3), this computed as

SE = M2(C + 1)(C – R) (5)

(R + 1)2 (R + 2)

And for our example above,

SE = 2002(250 + 1)(250 – 50) = 121.8

(50 + 1)2 (50 + 2)

The precision with which the capture estimates population size is

inversely dependent on the number of marked animals recaptured.

Thus, attempt to obtain a reasonably large R by making C large.

Confidence interval for mark recapture estimates may be

approximated from the standard error. A 1 – α confidence

interval may be computed as

N + (t)(SE) (6)

Where t is student’s t for DF = ∞ (i.e., a 95% confidence interval

calculation would use t = 1.96, and one would use t = 2.58 for a 99%

confidence interval).

12
For our above example, the 95% confidence interval for Equation
(2) could be calculated as:

988 + (1.96)(105.8) = 988 + 207

and we would say, with 95% confidence, that the population size
is between 781 and 1195.

The 95% confidence interval for Equation (3) could be calculated

as:

984 + (1.96)(121.8) = 984 + 238

and we would say, with 95% confidence, that the true population
size is between 746 and 1222.

Many modifications have been proposed to correct for some of

the limitations of the Lincoln-Petersen technique. Discussions of a

large number of them are in Andrewartha (1971). Burnham et al.

(1987), Caughley (1977), Krebs (1989), Pollack (1991), Pollack et

al. (1990), Seber (1982, 1986), and Southwood (1978).

Commonly recommended alternatives are the Schnabel method

and the Jolly-Seber method. Newer procedures often use theory

and sampling procedures based on multiple markings and/or

multiple recaptures. While such procedures may in some

13
circumstances yield more accurate estimates than the Lincoln-

Petersen procedure, they typically require more time and effort

(and they may have more restrictive assumptions). Some of these

methods are applicable if death, migration, or emigration alters

the ratio of marked to unmarked animals.

Assumptions of the Petersen Method

1. The population is closed, so N is constant

2. All animals have the same chance of getting caught in the first

sample

3. Marking individuals does not affect their catchability

4. Animals do not lose marks between the two sampling periods

5. All marks are reported on discovery in the second sample

Application

Capture-recapture methods may be applied to a variety of

animals. Field sampling may be done with sweep nets for larger

and slower arthropods or dip nets for benthic sampling of aquatic

invertebrates. Collect clams and other relatively sessile animals in

a small body of water by hand, employing random plots or

transects. Capture fish or crayfish with seines.

14
Small mammals such as brown rat may be live-trapped using a grid

of 100 traps spaced 10 to 20 m apart. Be sure to use the same

sampling effort on all sampling days. In the laboratory,

populations such as tenebrionid beetles, goldfish, or other

convenient animals, as well as inanimate objects, such as beans,

corks, or marbles, may be used to demonstrate the principles of

capture-recapture.

Inanimate objects may also be subjected to experiments to test

the importance of the assumptions given above and to investigate

the effect of sample size on the standard error of the population

estimate.

Identify, mark, record, and release animals as soon after capture

as possible. Animals with an exoskeleton or shell may be marked

with rapid-drying, weatherproof paint; dyes or more permanent

tags or bands may be applied to vertebrates. Fish may be marked

by clipping off a portion of the dorsal, caudal, or anal fin.

Amphibians or mammals may be toe-clipped.

Consider the permanence of the marking procedure. Markings on

exoskeletons or feathers will be lost if molting occurs between

sampling; also, some types of tags can wear off. The proper

15
elapsed time between samplings depends largely on how long it

takes for members of the first sample to distribute themselves

randomly within the entire population. A week or two should

suffice for most animals, but slow-moving forms may take longer.

Data in addition to numbers (e.g., weight, length, may also be

collected from these samples). To obtain an estimate of density
(D), an estimate of the area (A) sampled must be had in addition
to the estimate of population size (N):

D = N/A (7)

Indices. Indices are estimates of animal populations derived

from counts of animals signs, calls, roadside counts, and so on. In

this type of estimating all data are relative and must be

compared with data from other areas or other time periods. The

results do not give estimates of absolute populations, but they do

indicate trends of populations from year to year and from habitat

to habitat. Often this type of information is all that is needed.

16
Call Counts

Call counts are used chiefly to obtain population trends of certain

game birds, such as the mourning dove, bobwhite quail, woodcock,

and pheasant. A predetermined route is established along country

roads; it should be no longer than can be covered in one hour’s

time. Stations are located at quarter- or half-kilometer intervals,

depending upon the terrain, and the species involved. The exact

time to start must be determined for each area by the

investigator. The observer stops at each station, listens for a

standardized period of time (a minute or two), records the calls

heard, and goes on to the next station. Routes should be run

several times and an average taken. The number of calls divided

by the number of stops gives a call-index figure.

Roadside Count

The roadside count is similar to a call count, except with the

exception that the number of animals observed along the route is

recorded and the results divided by number of miles or

kilometers. Other variations include counting of animal tracks,

browse, signs, active dens and lodges, and so on.

17