Suburban Deer Population Model
Model development: This model is the result of an
adaptive management approach to estimating deer population
densities in suburban Chicago, Illinois. Empirical data
was combined in Stella 5.0 software (HPS 1997) to produce
deer population estimates in semi-isolated suburban landscape.
Empirical data was collected from a 1,500 km2 study
area encompassing Cook and DuPage Counties in northeastern
Illinois from winter 1992 through summer 1998. This data
includes survival and movement information from 147 radio-collared
white-tailed deer (129 females, 18 males) and sex ratio
and recruitment information collected from 2,599 lethally
removed deer (1,573 females, 1,026 males). Capture and culling
techniques were reviewed and approved by the University
of Illinois' Lab Animal Care Committee.
What makes this model unique is that it was constructed
from an extensive radio-telemetry data set for suburban
deer life histories used in combination with reproductive
data collected from continuous intensive annual deer culling
programs (1992 through 1997). This allowed us to build important
density-dependent factors directly into the model. Furthermore,
this model provides a tool for managers attempting to estimate
variable populations of suburban deer and provides credibility
founded in scientific research to highly scrutinized suburban
deer management programs. We thank the Forest Preserve Districts
of Cook and DuPage Counties, Cook County Animal Control
and Chicago Wilderness for funding this project.
Minimum system requirements:
Model input: The model is preset to run on an annual
increment beginning April 1 (time t) and ending April 1
the following year (time t +1). The model requires only
an initial population size and size of the area to be managed
(in km2). However, the user must keep in mind
that this is an accounting model designed for an adaptive
management approach and the more empirical data provided
by the user the better the predicted model output. Likewise,
the initial population estimate is crucial for accurately
depicting population estimates at time t + 1. We recommend
the best available data for this input. Initial
population estimates might be derived from adjusted aerial
or spotlight counts (Beringer et al. 1998, Farfarman and
DeYoung 1986) or if marked deer are available a capture-recapture
model design could be used (Pollock et al. 1990). The adult
deer sex ratio is also an important input, which can be
provided by the user. The proportion of females in the population
at time t has a significant influence within the density-dependent
recruitment function of the model and controlled management
programs tend to target mature females resulting in sexually
skewed populations. Previous harvest information or sex
ratios collected during deer counts could be applicable
for sex ratio input. We will discuss later how to incorporate
the previous years sex ratio harvest information into the
model. Manual removals (e.g., translocation or lethal removals)
also are input provided by the user. These are the removals
required to achieve a certain population size and can be
entered as a range of numbers using the sensitivity analysis
function in Stella 5.0. We encourage additional data input
from managers including survival information, immigration/emigration
rates or recruitment curves using the skeletal structure
of this model.
Model output: The model will provide an estimate
of deer population size (number of deer) or density
(deer per km2) at both monthly and annual
increments. The sensitivity-analysis function in Stella
5.0 software allows the user to predict annual population
shifts in response to harvest of females and males.
Some examples of questions the model was designed to
answer include, 1) given a known suburban deer population
size at time t what would be the population size at
time t +1 under different management strategies (e.g.,
no harvest, harvest of variable numbers deer, etc.),
2) given a population at a selected deer density at
time t what will be the predicted outcome of a set harvest
at time t +1, t + 2, etc.
Running the model: To run the model you first
need to enter the size of the area you intend to manage
and the initial population size. Double click on the
AREA symbol located in the bottom left corner of the
model. Enter the area (Km2) and click OK.
Now double click the INT POP box and type in a beginning
population number at time t (model preset at April 1)
and click OK. To run the model click RUN from the RUN
drop down menu. To review output double click on TABLE
1 (upper left hand corner of model). The output displays
a final population size and deer density at time t +1
(April 1 of the following year). To review output for
a different month, let's assume December 1, the model
must be adjusted. There are numerous ways to achieve
this, but the simplest is to stop the FINAL POP and
DEER DENSITY accounting functions at November 30. Remember
that the model is preset to run from time t (April 1,
month 1) to time t + 1 (April 1, month 13), so in this
time schedule December is month number 9. Therefore,
all additions and subtractions from the population must
be included through November 30th, but must
cease after this time. To achieve this first click the
X in the upper right corner of the screen to return
to the model and then double click on FINAL POP (bottom
center of model). Within the IF TIME statement change
the month (TIME) from 13 to 9 and select OK. Re-run
the model and view the output in TABLE 1 for FINAL POP
and DEER DENSITY. Observe how these numbers have changed
and appear next to month 8 accounting for all additions
and subtractions through November 30th. These
results are the December 1 (month 9) population estimate.
Now exit from the table and repeat this process changing
the month back to 13 in FINAL POP.
We discussed earlier the importance of adult deer sex
ratio in calculating the number of recruits into the
population. Sex ratio information is difficult to collect
in the field, but if deer are harvested in an unbiased
manner, it is acceptable to use the previous years harvest
ratio in an adaptive management approach. The present
sex ratio is set at 60:40 females to males, but this
can be changed if a different ratio is desired. This
is easily done by first double clicking on MALE RATE
located under INT POP in the center of the model. Then
simply enter the new rate and select OK. Repeat this
process for the FEMALE RATE so that the two combined
equal one-hundred percent. Also, a function built into
the model will automatically calculate the new sex ratio
after deer have been harvested from the population.
To calculate this first double click on FEMALES REMOVED
located in the upper right hand corner of the model.
Then enter the number of females harvested or estimated
to be harvested and select OK. Now do the same for MALES
REMOVED. Now double click on TABLE 1 and then double
click on the table again to reveal a selection menu.
Then, under the ALLOWABLE menu scroll down and select
NEW FEMALE RATIO by clicking to highlight and then clicking
the >> sign to the right and selecting OK. Now
run the model to review the output under NEW FEMALE
RATIO. This new ratio can then be inserted into FEMALE
RATE as described above. Also, be sure to edit MALE
RATE so that the two combined equal one-hundred percent.
Now when the model is run this new sex ratio will be
used in all calculations.
The strength of this model for management decisions
lies within the Sensitivity Analysis function of Stella
5.0, which allows for input of multiple simultaneous
deer removal scenarios. Exit from the table and select
SENSI-SPECS under the Run menu. Then click in the box
under # OF RUNS. Within the box enter the number of
different removal strategies you are proposing (e.g.
5 runs). Then, under the ALLOWABLE menu scroll down
and select first FEMALES REMOVED and then MALES REMOVED
by highlighting each and then clicking the >>
sign to the right. Then click on FEMLAES REMOVED under
SELECTED (VALUE) so that it is highlighted. Now click
on AD HOC under VARIATION TYPE and in the box to the
right of the AD HOC VALUES type the first number of
female deer you are predicting to remove from the population
(e.g., 0), and click set. This selected value will appear
in the box to the right under RUN #1. Now type in the
next value (e.g. 10) and select SET. Continue to enter
numbers until all RUNS #s have a corresponding VALUE.
Now click on MALES REMOVED from the SELECTED (VALUE)
menu and repeat the process. Once you have set your
runs for females and males click OK to return to the
model. Move the hand to TABLE 1 and double click. Double
click again on the table to get a selection screen.
Select COMPARATIVE under TABLE TYPE. Now select only
one (1) of the fields (usually FINAL POP or DEER DENSITY)
within the ALLOWABLE box by double clicking. This setup
will produce a comparison of the multiple model runs
for the selected variable. Make sure the ORIENTATION
(Vertical) and REPORT INTERVAL (Every DT) are checked.
Also, check ENDING BALANCES and INSTANTANEOUS under
REPORT and REPORT FLOW VALUES. When this is complete
click OK. Now select RUN from the drop down menu and
click S-RUN. TABLE 1 now displays the results of various
runs. Compare the runs and select a few which produced
desired population estimates or densities. Then go back
to SENSI-SPECS from the RUN drop down menu and click
FEMALES REMOVED. Scroll down the RUN # VALUE box to
see which numbers corresponded to the runs you selected
as good. Now do the same for MALES REMOVED. To refine
these numbers re-run SENSI-SPECS using values within
the range of good values chosen. To exit from S-RUN
and return the model to a standard RUN mode, double
click on the table and uncheck COMPARATIVE under TABLE
TYPE. Then reselect the variables you wish to view in
the table. Also, go back to SENSI-SPECS under the RUN
drop down menu and remove MALES and FEMALES REMOVED
from the SELECTED VALUE.
Most other model variables such as natural removal
rates (mortality rates) and the density-dependent recruitment
can be adjusted within their respective control variables
(see Stella 5.0 Technical Documentation).
Model verification: We recommend using an independent
variable for verifying model predictions whenever possible.
These variables might include deer-vehicle collisions,
count data collected independent of model input data
or harvest information. Verifying model predictions
will provide support for population estimates and help
direct possible input data adjustments.
Download the model
(6 KB 'zip' archive)
Beringer, J., L.P. Hansen and O. Sexton. 1998. Detection
rates of white-tailed deer with a helicopter over snow.
Wild. Soc. Bull. 26:24-28.
Farfarman, K.R. and C.A. DeYoung. 1986. Evaluation of
spotlight counts of deer in South Texas. Wildl. Soc. Bull.
Performance Systems, Inc. 1997. STELLA Research software.
Pollock, K.H., J.D. Nichols, C. Brownie and J.E. Hines.
1990. Statistical inference for capture-recapture experiments.
Wildl. Mono. 107:97pp
Starfield, A.M. 197. A pragmatic aproach to modeling for
wildlife management. J. Wildl. Manage. 61:261-270.
Questions or comments are welcome and should
be directed to the models authors:
Dwayne R. Etter, Timothy R. Van Deelen
Illinois Natural History Survey
607 East Peabody Drive
Champaign, IL 61820
Phone: (217) 333-6855
Fax: (217) 265-0374