BIOL 475 Lab 3: Estimating Population
Size Fall
2009
Introduction
The
size of a population is one of its most important attributes. Population size has critical
consequences for potential rates of evolution (gene flow, founder effects,
etc.) and vulnerability to extinction. The ability to measure population size
strongly influences our efforts to assess the impacts of various environmental
factors (abiotic, biotic, natural or
anthropogenic). In addition,
knowing the relative abundance of species in a habitat can give us some
important information about community structure. Finally, observing the changes
in populations of organisms through time yields insight into the natural
variability inherent in populations and the ability to identify trends through
time in response to external forcing agents. Population size can be expressed
as the absolute number of organisms in an area under consideration (such as the
number of squirrels on the College campus), or, alternatively as population density (the number of
squirrels per unit area, such as 2 squirrels per tree or 20 squirrels per 100m2).
Although
it is extremely important to know the population size of the organism(s) under
study, measuring it presents a number of problems. First, organisms differ in their
level of detectability (how easy it is to find them),
either because of their behavior, size, coloration, or habitat. In general, in
natural systems, it would not be practical to attempt to count every individual
in the system. Even when the U.S. government spends tens of millions of dollars
to conduct a national census to estimate the size (and character) of the human
population, the raw counts are often not accurate. Ecologists usually need to
gather this kind of information with tens of dollars. Therefore, we must resort
to estimating population size through sampling. There are a variety of
different methods to do this, each with associated assumptions. The best
estimates are those for which you can calculate a measure of certainty—a
confidence interval—to give you an idea of how accurate your estimate is.
Ideally, we use a 95% confidence interval, which identifies the spread, or
range of values between which we are reasonably certain lies the ÔtrueÕ
population size. If we were to sample this same population again and again,
each time calculating an estimated population density, then 95% of these
samples would estimate the population size to be within the confidence
interval. This gives us an indication of how close we probably are to a real
measure of population size. It tells us how ÒgoodÓ our estimate is. In general,
the larger the sample size we take, the better is our estimate of true
population size. In designing a
sampling method for collecting data, we try to use a sample size that gives us
a reasonably good estimate of population size but which minimizes the effort or
expense needed to make the sample.
Because
there are so many different kinds of organisms, each with its own behavior and
natural history, estimating the size of a population in nature takes planning,
effort, and ingenuity. It also requires that you know something about the basic
biology of the organism you are measuring. Different sampling methods are
appropriate for organisms that are stationary
(such as plants or intertidal invertebrates) and for those that move (most terrestrial and aquatic
animals). For stationary organisms, sampling usually involves the use of
permanent quadrats, or sampling enclosures, which are
randomly placed in the population, and within which the number of organisms is
then counted at intervals. Mobile
organisms can also be censused by using sampling quadrats, by trapping animals, by spending a certain
constant amount of time searching through the habitat, or by various other
means. Once again, the main goal of an ecologist is to determine the most
accurate method to use that best suits the organism and the resources available to the ecologist.
In this
weekÕs lab, we are going to estimate the population size of grasshoppers
inhabiting an old field. Grasshoppers will be our focal organism because they
are present in high numbers, are relatively easy to catch, are unlikely to
leave the field in three days and they donÕt bite or sting! The old field we
will be sampling in is a habitat island, in that it is surrounded on all sides
by areas unlikely to be appealing to grasshoppers. We are going to use and evaluate
two different methods to estimate population size: mark-recapture and
depletion, sometimes known as Ôcapture per unit effortÕ. I guarantee that assumptions
will be violated! But, the key in ecology is to violate as few assumptions as
possible, while analyzing how these violations might affect the outcome.
Methods: Estimating Population Size in Mobile Organisms
I. Mark-Recapture Method
The
simplest form of this method is called the Lincoln-Peterson
Index. It consists of taking a random sample of individuals from a
population, marking the individuals, and releasing them back into the
population. After an appropriate amount of time (which will depend on the
biology of the organism), a second sample is taken and the number of marked and
unmarked individuals in the second sample is counted. If certain assumptions
are met, then the proportion of marked individuals in the second sample is an
estimate of the percentage of the total population that comprised the initial
sample. For this method to yield a reliable estimate, the following assumptions
must be met:
a)
Marked and
unmarked individuals mix randomly after the first capture
b)
The population
is closed (no emigration or immigration & no births or deaths)
c)
All individuals
have an equal probability of capture
d)
All sampling is
done at random.
e)
Being captured
once does not affect the probability of being captured again later.
II. Depletion Method
The
depletion method is based on the idea of diminishing returns, and makes the
same assumptions as the mark-recapture technique. If successive collections are
made from a population and the individuals collected are not returned to the population, then a gradual decrease in the
population size will result as individuals are removed. If the same effort is made to capture organisms
during each collecting period, then a gradual decrease in the number collected
will also occur. If the number caught during each sampling period is plotted as
a function of the cumulative number collected, then the points theoretically
should fall along a straight line (see below). We can use this line to predict
how many individuals would have been accumulated had we been able to capture
every one. The point on the line where the catch per unit effort equals 0
corresponds to the value of X where all members of the population have been
captured, i.e., the estimated population size, N. This is shown on the
following graph:

Catch
per unit effort plotted as a function of the
cumulative number collected prior to the current sampling
period.
Sampling Procedure
Although
we will calculate our estimates of the grasshopper population size separately
for each method, we will partially combine the sampling effort to maximize our
sampling efficiency (i.e., get the largest sample in the shortest amount of
time). To accomplish this, we will mark off two sets of three 5 X 5 meter
sampling quadrats. The locations of these quadrats will be randomly determined. Once the locations
are decided on, a barrier to immigration and emigration will be constructed
using 8 bamboo polls, bridal veiling, and a few binding clips (if you canÕt do
an ecological study with materials bought at exclusively at a discount store,
then youÕre doing something wrong). Once the structure is complete, students in
each group should identify a pair of grasshopper catchers, a data-logger, and a
team of grasshopper markers.
For
sampling done within the quadrats as part of the
depletion method, the unit of effort we will be using for this exercise is hand
collection (with optional butterfly nets) over 15 minute intervals. During this
time, all grasshoppers encountered by either of the two collectors will be
caught and placed in a ziplock bag. At the end of the
15 minutes, collection will cease and the collectors will hand the bags to the
data-logger, who will count and record the number of
grasshoppers, and pass them to the marking team. The marking team will
carefully transfer the grasshoppers to a marking chamber (plastic box) and
apply a touch of white-out to the thorax of the
animal. Marked grasshoppers will be placed back into their original ziplock bags. This procedure will be repeated a minimum of
4 times. Once the four collections have been collected, counted and marked, we
will dismantle the enclosures and release the marked individuals back into
randomly selected locations in the field.
As
a class, we will conduct an immediate recapture effort once the grasshoppers
are released. Students will spread out and catch as many grasshoppers as we can
in the time we have. Be careful not to TRY and get marked individuals; instead,
go for whatever individuals you come across. Our goal is to measure the proportion of individuals in our sample
that are marked. I will attempt to repeat this the recapture on Monday
afternoon if I can round up some help. The justification for doing the
recapture twice is that we will most likely be violating assumption (b) stated
above. Although I chose this field because it is surrounded by unlike vegetation and
so is effectively a somewhat closed system, we cannot prevent deaths from
occurring over the course of even 2 days (i.e., some of these grasshoppers will
be eaten by SOMETHING). In this case, we may be violating assumption (a), as
the marked grasshoppers may not be randomly dispersed among the population.
Still, the difference between these two recapture attempts should tell us some
interesting things.
Estimating Population Size
I. Mark-Recapture. Because (hopefully) two separate recapture attempts
were performed, we will have two separate population size estimates. Each
should be calculated as follows.
The
raw data for the calculations are as follows:
M =
number of individuals marked and released in the initial sample.
C =
total number of individuals captured in the second sample.
R =
number of marked individuals recaptured in the second sample.
The
population estimate is based on the assumption that, if we mark a certain proportion
of the population, we should recover an equivalent fraction of marked animals
at a later sampling time. If, at the second sampling interval, 12% of the
animals we capture are marked, and if (as the assumptions above state) every
animal is equally likely to be captured, then we can assume that, in our first
census, we marked that same fraction of the total population.
M/N
= R/C
By
rearranging this equation, we obtain the Lincoln-Peterson Index, which gives us
an
estimate of
population size (N):
N
= MC/R
Confidence
intervals for N are calculated as follows. Let us define the following
variables:
p =
decimal fraction of marked animals
that were in the second sample.
=
R/C
q =
decimal fraction of unmarked animals that
were in the second sample.
=
1-p
Then,
the 95% Confidence Interval is expressed as:
95%
C.I. = p ± 1.96
We
can say, therefore, that this interval will include the true proportion of
marked individuals 95% of the time.
( p - 1.96 ) and (
p + 1.96
)
Similarly,
we can be 95% sure that the actual population size lies within the range,
(N
- 1.96 N) and ( N + 1.96 N)
The
four values you obtained by sampling (the catch
per unit effort) are to be plotted as Y values. The cumulative numbers of grasshoppers collected in all the samples up
until that time (i.e, the total number of
grasshoppers removed from the population in all the samples prior to the
current sample) are to be plotted as X values. In the first sample, Y will be
the number of grasshoppers you sampled in the first 15 minute
increment, and X will be zero (since, prior to this sample, no grasshoppers had
yet been removed from the population). In sample 2, Y will again be the number
captured in that second sample interval, and X will be the number of
grasshoppers removed from the population in all the previous samples (in this
instance, those taken at sample 1). Theoretically,
we expect the values of Y to decrease linearly as X increases. This means,
biologically, that as we remove more and more individuals from the population
(X increases), the number of grasshoppers we encounter in each successive unit
effort should decline (Y decreases; i.e., the points should lie along a
straight line). However, because
of sampling error, the relationship might not be exactly a straight line; there
will be error around the predicted line. Plot the data points for each of the
study quadrats, fit a straight line to the data and
calculate the x intercept when y = 0 (see figure above). [hint:
use Excel to fit a line to the data. Then, once you have the equation for the
line, you can solve for x] Biologically, this method says that, when we have
removed so many organisms from the population that we do not capture any more
in a sampling interval, we have essentially collected all the individuals in
the population- and this is, by definition, the population size. So when the
catch per unit effort is 0, the cumulative prior catch is an estimate of N.
Np = (NsRp)/Rs
Where
Ns is the number of estimated individuals
in our study plot, Rp is the range of the
population (size of the field) and Rs is
the area of the study plot (25 square meters).
Once
you have calculated the six estimates of the population size of grasshoppers in
this field, calculate the mean and the 95% confidence interval about the mean
[hint: the 95% confidence interval in this situation is calculated as 1.96*(the
standard deviation)].
Assignment due Friday October 2
1. Calculate the Lincoln-Peterson Index using the class
recapture data (to be posted by Tuesday morning) along with the 95% confidence
intervals. This will give you an estimate of the number of grasshoppers in the
field. Based on the characteristics of the field we sampled in, which
assumptions do you think were violated most significantly and what effect do
you think this might have on our population estimate?
2. Calculate Np
using the Depletion Method described above based on the data from ONE quadrat. Which assumptions of this method do you think were
violated most significantly in our sample and what effect do you think this might
have on our population estimate?
3. Which method do you think gives the most accurate
estimate of population size for this particular field?