Wildlife Society Bulletin 21:96-97, 1993



                             WILDLIFE SOFTWARE



                    COUNT--A SIMULATION FOR LEARNING TO

                         ESTIMATE WILDLIFE NUMBERS



JOHN I. HODGES, Wildlife Counts, 318 Coleman Street, Juneau, AK 99801



     Population surveys are a major part of wildlife biology and wildlife management.

Some surveys involve large numbers of animals where observers have a limited amount

of time in which to make their count. Estimates must be used in lieu of exact numbers.

Examples include waterfowl flocks, sea bird colonies, shorebird flights, caribou herds,

and salmon runs. Inaccuracies from estimation often result in large variances and

observer bias (Erwin 1982). Correction factors vary by observer, but are usually

illusive because the actual number of animals is difficult to measure (Samuel and

Pollock 1981). Observer training and standardization may reduce these problems, and

traditionally have been accomplished by using photographs of animals or by throwing

handfuls of beans on the floor. Animals in each photograph must be counted, either by

hand or more recently by a computer, if the images sufficiently contrast with the

background (Bajzak and Piatt 1990). Survey managers often ignore this critical aspect

of ocular surveys when using inexperienced observers or replacing long-standing

observers.

     Procedures for dealing with the problem of not seeing all of the individuals on

a survey have been extensively studied (Pollock and Kendall 1987). Little treatment

has been given to the problem of estimating large numbers of individuals when all

individuals are quite visible. Caughley et al. (1976:300) recognized the need to deal

with shortcomings of observer estimation, and stated ". . . a further difficulty is

introduced by the observers- inability to tally high numbers quickly (counting

saturation) which gives rise to a completely new set of biases." Whether population

surveys are accomplished with line transects, plots, or complete censuses, they often

require observers to estimate large numbers of animals in groups. Sample variance and

animal detection problems may be the primary sources of error in some surveys whereas

observer estimation problems may be the primary source of error in others. In the

latter case, observer training may reduce observer error. However, training flights

are an expensive way to practice and the observer does not know if his or her estimates

are accurate. In much the same way that flight simulators are used to train pilots for

flying, computer simulations may be used to train observers to estimate wildlife

populations.

     The computer program COUNT is a tool for training observers who estimate

wildlife and fish populations. The number of images is chosen at random and displayed

in 1 of 6 different settings. SWANS ON A LAKE gives the perspective of distance.

FLUSHED GEESE provides the confusion of beating wings. DUCKS FROM AN AIRPLANE

distributes ducks within the irregular boundaries of imaginary ponds. RANDOM POINTS

simulates configurations of shorebirds. SALMON IN A CREEK is viewed from an overhead

perspective. OVERFLIGHT OF BIRDS scrolls flocks of birds across the screen to simulate

migration.

     The observer may adjust the viewing time allotted, and place upper and lower

bounds on the number of images. The maximum number of individuals is 5,000 in

OVERFLIGHT OF BIRDS. At the end of the chosen number of trials, the results are

displayed showing the observer's estimate, the actual number of images, and the percent

error by trial. All trials are completely random unless the same series of displays

is desired, in which case a common random seed may be used.

     Professional wildlife biologists and college students are obvious beneficiaries

of this software, but students of all ages also can enjoy the challenge of learning

to estimate large numbers of animals or dots. Bird-watchers who participate in formal

surveys need this type of training as well. Friendly competition between users often

results when the true number of birds displayed is known.

     The program is written in QUICK BASIC 4.0 for IBM-compatible computers. Color

VGA or EGA monitors give superior graphics but the program can use black and white

monitors with graphics capability. A version for Apple II computers also is available.

It is simple to run and suited to classroom use. An information brochure on the program

is available from Wildlife Counts, 2215 Meadow Lane, Juneau AK 99801. Cost of the

program currently is $49.





                             LITERATURE CITED



BAJZAK, D., AND J. E. PIATT.1990. Computer-aide procedure for counting waterfowl on

     aerial photographs. Wildl. Soc. Bull. 18:125-129.



CAUGHLEY, G., R. SINCLAIR, AND D. SCOTT-KEMMIS. 1976. Experiments in aerial survey J.

     Wildl.. Manage. 40:290-300.



ERWIN, M. R. 1982. Observer variability in estimating numbers: an experiment. J Field

     Ornithol. 53:159-167.



POLLOCK, K. H., AND W. L. KENDALL. 1987. Visibility bias in aerial surveys: a review

     Of estimation procedures J. Wildl.. Manage. 51:502-510.



SAMUEL, M. D., AND K. H. POLLOCK. 1981. Correction Of visibility bias in aerial surveys

     where animals occur in groups. J Wildl. Manage. 45:993-997.



Received 20 December 1991. 

Accepted 26 August 1992. 

Software Editor: Rexstad.