Isolation of total genomic DNA was accomplished by extraction from freshly ground flies and purification by CsCl density gradient centrifugation (Bingham et al., 1981). Genomic DNA was digested with HindIII and EcoRI restriction enzymes and then loaded onto a 5–30% sucrose step gradient (Ausubel et al., 1987). The fraction containing 5- to 6-kb fragments was cloned into the vector pBluescript SK—(Stratagene) and transformed by high-voltage electroporation (Maniatis et al., 1982, Dower et al., 1988). D. pseudoobscura clones containing Amy homologous sequences were isolated from a genomic library by colony hybridization (Maniatis et al., 1982) using the plasmid pFA4 (Brown et al., 1990) containing the D. pseudoobscura Amy1 coding sequence as probe. All sequences were determined using an automatic sequencer (Applied Biosystems 373A) following the manufacturer's protocol. In all cases both strands were sequenced. Some regions were sequenced again manually (Sanger et al., 1977) using the Sequenase DNA sequencing kit (United States Biochemical). Sequenced regions include 667 nucleotides (from bases 701 to 35) upstream of the start codon and 391 nucleotides (from bases 62 to 452) downstream of the stop codon. These sequences have been deposited in GenBank (accession numbers U09746–U09757). Sequences were aligned with each other using the GENALIGN program (IntelliGenetics) and checked again by eye. Sequence divergence estimates were calculated as direct counts of nucleotide sequence differences, since no correction is needed for differences as small as those in our study (Nei, 1987).
The phylogenetic analysis was carried out using the neighbor-joining (NJ) and maximum likelihood (ML) methods in the PHYLIP package (Felsenstein, 1993), the NJ method in the MEGA package (Kumar et al., 1993), and the maximum parsimony (MP) method in PAUP (Swofford, 1991). For the RSP data, we excluded strains with a redundant restriction pattern, which reduced the number of strains that were phylogenetically analyzed from 33 to 21. To bootstrap the NJ tree derived from RSP data, we followed the advice kindly given to us by Walter Fitch. First, "1" and "0" in the data set were replaced with "A" and "T," respectively. Second, the SEQBOOT program (PHYLIP) was used to produce 100 bootstrapped data sets. Third, the DNADIST, NEIGHBOR, and CONSENSE programs (PHYLIP) were used in succession to produce the bootstrap values.
A Test of Hypotheses Regarding the Ancestral Gene Arrangement. Amylase in D. pseudoobscura is a family of three genes, located within the inverted region in most gene arrangements (Aquadro et al., 1991). The Amy1 gene