RESULTS
Table 1 shows the frequencies of both midgut patterns and Amy alleles in the four populations over a period of 800 days. The sample size for the midguts is between 60 and 66 for each cage at each sample; 96 genes were sampled for Amy for each cage. These populations have only two Amy alleles which we designate F and S and almost certainly correspond to PRAKASH and LEWONTIN'S (1 968) 1.00 and 0.84 alleles, respectively. There is no pattern of Amy allele changes evident. Even after more than 2 yr all four cages have very similar gene frequencies, which are not significantly different from the original frequencies. Thus, there is no evidence for selection on these alleles. We might also note that these results give no evidence for drift being sufficient. This adds to our confidence that such experimental populations have rather large effective population sizes, and bottlenecks are not common. The changes in midgut patterns are significant and somewhat complex. Because the sample size is small for several patterns, statistical tests on the raw data are suspect. Thus, we have combined categories for testing. In essence, we have ignored the PMG and combined categories with the same AMG pattern. Thus, we have four categories, A + E, B + F, C + G and D + H (lettered designations as in Table 1). A FUNCAT analysis (SAS Institute, Cary, North Carolina) was performed on the combined data; this is similar to an analysis of variance based on chi-squares.
Table 2 presents the results.
There are significant cage effects indicating that the frequencies are different in different cages; the cage X time effect indicates that the cage differences were not static but changed over the course of the experiment. When the data in Table 1 are examined, one can see some patterns in these significant changes. In the starch cages the frequency of AMG pattern 100 (regardless of PMG) has increased, whereas in the maltose cages it has decreased. In maltose cages, AMG 123 has increased to 27% by day 800, whereas in the starch cages, it has remained 8 and 14%.
DISCUSSION
The general conclusion is that there is a tendency for limiting amylase activity to the most anterior region of the AMG in starch cages, whereas amylase activity spread more evenly along the midgut is favored in maltose cages, regardless of what is happening in the PMG. This result is similar to that obtained previously when a different base population (Bryce) was used (POWELL and ANDJELKOVIC 1983). In that study, the pattern 100-00 increased rather dramatically in the starch cages, whereas little or no change occurred in maltose cages. The main difference is that, in the present study, the PMG pattern did not seem to respond as it did with the Bryce populations. Based on the four populations studied in the present report, we would not be overly confident of the conclusions stated. However, because they are similar to those obtained with 14 other cages, we feel that they are quite significant. Thus, a total of 18 cage populations have been studied for changes in midgut activity patterns, using two different base populations. The robustness of the conclusion (that selection affects this polymorphism differently in starch and maltose environments) is established. In contrast, the results on the frequency of Amy alleles are not so clear. Previously (POWELL and ANDJELKOVIC 1983), we did obtain evidence of selection on this polymorphism as well. However, there is no evidence of selection in the present data. We can only conclude that the “genetic background” can affect the outcome of selection at Amy. An alternative explanation is that the F and S alleles are not really the same in the two populations. However, neither POWELL (1979) nor NORMAN (1978) found “hidden” alleles at Amy of D. pseudoobscuru by varying pH and acrylamide concentration. Thus, this is not a likely explanation, and we are left with genetic background as the most likely factor. Such effects could be due to either close linkage and hitchhiking or to epistasis; it would be difficult to distinguish between these. That the two kinds of polymorphism respond differently is not particularly surprising. Our previous studies on natural populations led us to conclude that the evolutionary dynamics of the two systems are independent. This was based on the fact that there is no correlation between the changes in morph frequencies of the two systems from population to population within a species (POWELL 1979) or across different species (POWELL, RICO and ANJELKOVIC 1980). It would seem clear that selection can operate independently on a structural gene and the factors that control its expression. The rather extensive DNA sequence data being accumulated would seem consistent with this conclusion (see e.g., EFSTRATIADIS et al. 1980). The pattern and extent of changes within a coding region are usually quite different from changes in noncoding regions, some of which are assumed to be controlling regions. Our results are populational confirmation of these molecular observations.
