II.  Empirical analysis of genetic drift: Buri's studies on drift in lab populations of
      D. melanogaster

     A.  Behavior of ensembles of populations under genetic drift

         1.  Because genetic drift is a chance, statistical phenomenon, we can only characterize its properties by probabilities

         2.  Empirical analysis of whether gene frequencies behave under drift as predicted must therefore involve examining
              large samples of gene-fequency changes to estimate their statistical behavior.

         3.  Easiest way to do this is to characterize gene frequency changes in large ensembles of populations.

         4.  Predictions for a large ensemble of populations, all starting with the same initial frequency:

            a.  Frequency histogram of gene frequencies gradually widens (variance becomes larger).
            b.  Populations accumulate in the fixed classes (i.e. p = 0 or p = 1).
            c.  The relative proportions in the unfixed classes (corresponding to ) settle quickly to an equilibrium
            d.  All populations eventually become fixed for one allele or the other
            e.  The expected proportion of populations fixed for a particular allele is equal to the initial frequency of that allele.

     B.  Buri's Experiment: Overview

         1.  Examined gene frequency change at bw locus affecting eye-color

            a.  Two alleles: bw and bw⁷⁵
            b.  bw⁷⁵/bw⁷⁵ have bright orange eyes
                 bw⁷⁵/bw    have light orange eyes
                 bw/bw       have white eyes

         2.  Preliminary experiments: confirmed absence of detectable selection at this locus in laboratory
              environment--necessary to ensure that only genetic drift was operating at this locus

         3.  Main experiment: examine change over time in frequency histogram of gene frequency for 107 replicate
              laboratory populations.

     C.  Preliminary experiments

         1.  Examined whether observed proportions of genotypes from particular crosses deviated from expected
              proportions to verify absence of selection (see previous explanation of Measuring Fitness in Drosophila
              laboratory populations).

         2.  Results

            a.  In all but three tests, observed proportions did not deviate significantly from expected proportions.
            b.  Expect three significant deviations just by chance.
            c.  Repeated  three tests that showed significant deviations and found that there were not significant deviations
                 from expectation.
            d.  Conclusion:  no detectable selection acting on this locus in laboratory environment.

     D.  Main experiment

        1.  Procedure
 
            a.  Start 107 replicate populations with initial frequency of brown-eye allele of 0.5.
            b.  Let each population reproduce on its own.
            c.  Determine gene frequency in emerging adults.
            d.  From emerging adults, choose 8 males and 8 females randomly to found the next generation.  Remove
                 these individuals and place in new population cage.
            e.  Repeat steps b-d for 20 generations.

        2.  Results

            a.  Frequency histogram of gene frequencies spreads as predicted
            b.  The relative frequencies of the unfixed classes do not deviate from expectations

    E.  Conclusions: Genetic drift in these laboratory populations behaves according to theory.