Testing the Validity of the Lincoln Index Using a
Simulation
Aim
The aim of this investigation is to use a simple model to assess the accuracy with which the
Lincoln Index estimates population size and thus determine its validity in habitat studies.
Research Question
How accurate is the Lincoln index in predicting population size, based on a model using a known
number of seeds?
Results
Where:
n1 = first capture number
n2 = recapture number
m2 = marked individual in recapture
N = estimated population size
Test number
1 2 3 4 5 6
n1 32 13 18 14 9 8
n2 28 17 17 21 8 27
m2 7 2 3 6 1 1
N 128 110.5 102 49 92 216*
N Average 96.3
Range 49 – 128
*Value discounted as anomalous
Average value deviates from true value by 3.7%
Maximum error (49) deviates from true value by 51.0%
Analysis and Discussion
The true population size is known to be 100. The average population size estimated by this
method is 96.3, though the range is from 49 to 128. An outlying population size of 216 was
estimated in one of the repeats and this was considered to be anomalous and therefore discarded
in the data processing.
The size of the “capture” and “recapture” instances appears to very loosely correlate with the
accuracy of the population estimate:
60
Deviation from true population (%)
50
40
30
20
10
0
0 10 20 30 40 50 60 70
Sum of n1 and n2
Graph 1
Conclusion
Based on the maximum error in this investigation, it can be concluded that a population estimated
using the Lincoln Index can be considered accurate by plus/minus 51%. This suggests that the
Lincoln Index is very unreliable in estimating population sizes. However, when an average of
several repeats is created, the value is 3.7 % from the true value.
The inference is that the Lincoln Index should be repeated in ecological studies and a mean value
attained to ensure accuracy in the value.
Evaluation
Strengths of the simulation:
The simulation involves a small sample size of easy to count seeds. “Animals” (represented by the
seeds) can easily be fully mixed in the “ecosystem” (represented by the beaker). There is no
death, birth, immigration or emigration in the simulation. These are factors which reduce the
reliability of a Lincoln Index value produced in “real-world” scenarios and thus this simulation is
more effective for testing the validity (as opposed to using a known population of real animals in a
natural ecosystem). Similarly, the possibility of marking having an influence on the counts is nil; in
a “real-world” scenario, marking animals may influence the counts through toxicity of the markings,
enhanced predation etc.
Weaknesses and limitations of the simulation:
The simulation involves only a small sample size of 100. True population sizes in natural
ecosystems will range widely, possibly up to millions of individuals. This simulation do not attempt
to consider how valid the Lincoln index is at various population sizes.
The simulation involves conducting repeats and identifying anomalous counts to be discarded. In
reality, the Lincoln index counts would be created based on a single capture-mark-recapture
process. The inference here is that this simulation may estimate the Lincoln Index to be more
accurate than it truly is when used in real studies.
Suggestions for Improvement
1. The simulation should be repeated with larger sample sizes to more appropriately reflect
real uses of the Lincoln Index.
2. The simulation should be repeated with a range of known population sizes to determine
whether validity varies with population size.
3. More repeats should be conducted to establish likelihood of anomalous results being
generated.
4. The simulation should be repeated with a range of “capture” and “recapture” values to
determine whether this value has an influence on the reliability of the data, as is suggested
by these results (See graph 1). This could be achieved by using collection devices of
differing sizes such as a teaspoon, tablespoon, cup etc., which maintains the randomness
of sample size required in the simulation. If this is determined to have an influence on the
accuracy of the population estimate, it would suggest the size of trap and length of time
over which organisms are captured should be considered in real ecological studies.
