Luikart, Gordon, Jean-Marie Cornuet, and Fred W. Allendorf. 1999. Temporal
changes in allele frequencies provide estimates of population bottleneck size. Conservation
Biology 13: p.523-30.
(Amy Pedersen)
The authors utilized the temporal method to estimate Ne (effective population size).
They then examined the biases associated with this method, as well as the causes
of these biases, and attempted to correct for these biases.
Some basic assumptions were made before evaluations were made using Monte Carlo computer
simulations: the examined loci were selectively neutral, new mutations were not
incorporated (because any effect of mutations was considered negligible due to the
small population size and brevity of the study), and to acquire a suitable sample
size (30 individuals or 60 gametes), it was assumed that 30 offspring would result
from four individuals (Ne=4).
The authors analyzed the results of the Monte Carlo simulations and determined the
biases. They found that bias (specifically overestimation of Ne) increased with
the severity of the bottleneck, i.e. smaller Ne yielded greater overestimation.
However, when Ne=4 (a severe bottleneck), bias did not get "large enough to
cause bottlenecks to go undetected". Bias was found to be negligible for larger
estimates of Ne unless the bottleneck lasted for more than five generations.
It was uncovered that three factors brought about bias, namely the loss of variation
(alleles), "the assumption that the increase in variance in allele frequencies
with time is additive (instead of multiplicative), and the assumption that selfing
can occur when it cannot".
The authors found that the precision and power of the temporal method could be altered
when different parameters of the simulation were changed. They discovered that the
main limitation of the temporal method was that the upper confidence limits were
often too high when estimating Ne. However, suggestions (below) were made to remedy
this situation.
The authors made some recommendations to help alleviate the overestimation of Ne.
They derived a new equation for estimating Ne which assumed that selfing did not
occur and the increase in variance in allele frequencies was additive. This equation
significantly reduced the overestimation of Ne. A suggestion that corrected for
the overestimation of Ne caused by loss of alleles was to only monitor alleles that
were found to be above a certain frequency. This minimized the possibility that
the studied alleles would be lost in the first few generations of a bottleneck.
They also suggested frequent sampling (one-generation sampling interval) and increasing
the number of individuals or alleles sampled to increase precision of the estimation
of Ne. Other suggestions included development of the temporal method into a "more
formal statistical test" and further study of statistical analysis including
bootstrapping.