is affected as the harmonic mean of population sizes in each
<br />generation, or 1/Ne = 1/t (11N, + 11N, + ... + 1/N,),
<br />where t = time in generations (Franklin 1980). Comparison
<br />of the following two populations illustrates the importance
<br />of fluctuations to Ne. In the first, there are 100 fish for each
<br />of 5 consecutive generations, for an arithmetic mean of 100;
<br />1/Ne = 1/5 (1/100 + 1/100 + 1/100 + 1/100 + 1/100) =
<br />.01, and Ne = 100. In the second case, the arithmetic mean
<br />is also 100, but the population fluctuates each generation as
<br />follows: 100, 10, 300, 10, 80; 1/Ne = 1/5 (1/100 + 1/10 +
<br />1/300 + 1/10 + 1/80) = .045, and Ne = 22. In this case,
<br />Ne for five generations is reduced by 78% through popu-
<br />lation crashes.
<br />The importance of Ne to population genetic structure is
<br />immediately realized in consideration of three closely related
<br />problems of small populations: bottlenecks, drift, and in-
<br />breeding.
<br />Genetic Bottlenecks. A bottleneck is a sudden and dra-
<br />matic decline in numbers (Nei et al. 1975). Bottlenecks ef-
<br />fectively sample (althou&h not necessarily randomly) a few
<br />individuals from a larger gene pool, resulting in a remnant
<br />population with less overall variation. The degree to which
<br />the new population is genetically depauperate will depend
<br />at least on genetic diversity of the source population, size
<br />of the bottleneck, and the degree of randomness of selection
<br />of individuals.
<br />Loss of variation during a bottleneck has two components
<br />(Nei et al. 1975). First, is reduction in variance of quantitative
<br />traits. The proportion of quantitative variation remaining
<br />after a single bottleneck is approximately 1 - 1
<br />GN ,where
<br />N is the number of individuals surviving (Frankel and Soule
<br />1981). A small number of individuals contains most of the
<br />(D
<br />_Z
<br />Z_
<br />Q
<br />W
<br />fY
<br />W
<br />U
<br />Q
<br />tY
<br />Q
<br />o.
<br />POPULATION SIZE (Ne)
<br />Figure 1. Proportion of original genetic variance remaining in pop-
<br />ulations of various sizes after a single bottleneck.
<br />16
<br />genetic variation of a source population (Fig. 1): two indi-
<br />viduals contain 75% and 10 contain 95%. Therefore, unless
<br />a bottleneck is very severe or is prolonged over several gen-
<br />erations, it will not drastically reduce the amount of quan-
<br />titative variation.
<br />Second, and more critical, is loss of specific, and usually
<br />rare alleles in a bottleneck (Denniston 1978). Alleles at fre-
<br />quencies of, say, 5% or less,. contribute little to overall ge-
<br />netic variance, but may periodically be important to the
<br />population as a whole (Frankel and Sould 1981). Figure 2
<br />illustrates the effect of bottlenecks of various sizes on rare
<br />alleles, assuming a locus with initially six alleles at frequen-
<br />cies of .90, .02, .02, .02, .02 and .02. With a bottleneck pop-
<br />ulation size of 10, there would be an average loss of more
<br />than three alleles at this locus. Considering the repetition
<br />of this loss over hundreds or even thousands of similarly
<br />variable loci, such losses can be considerable.
<br />Genetic Drift. Genetic drift is random change in gene fre-
<br />quency due to sampling error in small populations (Li 1976).
<br />It is, in effect, a prolonged bottleneck leading to repeated
<br />loss of variance until, in its ultimate form, all loci are fixed,
<br />with complete absence of genetic variance. Even in the pres-
<br />ence of moderate selection pressures, drift can be a potent
<br />force in small populations (Wright 1931, 1948).
<br />The depletion-/of genetic variation through drift is esti-
<br />mated by l1 - I L t, where t is the number of genera-
<br />6
<br />Ln
<br />W 5
<br />J
<br />W
<br />J
<br />J 4
<br />Q
<br />O 3
<br />r'r
<br />W 2
<br />co
<br />Z
<br />POPULATION SIZE (Ne)
<br />Figure 2. Expected number of alleles present after a single bottle-
<br />neck in populations of various sizes, derived from a source pop-
<br />ulation with six alleles at frequencies of .90, .02, .02, .02, .02 and
<br />.02. Expected number of alleles remaining at a locus is calculated
<br />as E(n) = m - E (1 - p,)-,, where m = original number of alleles
<br />and p; is the frequency of the ith allele (from Denniston 1978)
<br />Fisheries, Vol. 11, No. 1
<br />1 10 100
<br />1 10 100
|