Flow from Rare Alleles

In order to assess the evolutionary significance of molecular variation in natural populations of Drosophila melanogaster, we have started a comprehensive genetic variation study program employing a relatively large number of gene-protein loci and an array of populations obtained from various geographic locations throughout the world. In this first report we provide estimates of gene flow based on the spatial distributions of rare alleles at 117 gene loci in 15 worldwide populations of D. melunogaster. Estimates of N m (number of migrants exchanged per generation among populations) range from 1.09 in East-Asian populations (Taiwan, Vietnam and Australia) to 2.66 in West-Coast populations of North America. These estimates, among geographic populations separated by hundreds or even thousands of miles, suggest that gene flow among neighboring populations of D. melanogaster is quite extensive. This means that, for selectively neutral genes, we should expect little differentiation among neighboring populations. A survey of eight West-Coast populations of D. melunogaster (geographically comparable to Drosophila pseudoobscura) showed that in spite of extensive gene flow, populations of D. melanogaster show much more geographic differentiation than comparable populations of D. pseudoobscura. From this we conclude that migration in combination with natural selection rather than migration alone is responsible for the geographic uniformity of molecular polymorphisms in D. pseudoobscura. HE role of gene flow in the genetic structure of T species has been a major source of arguments in evolutionary biology (MAYR 1963; EHRLICH and RAVEN 1969; ENDLER 1977; STANLEY 1979). Gene flow in combination with random genetic drift or natural selection determines the extent to which geographic populations can differentiate from one another. WRIGHT (1 93 1) showed that an island population of size N receiving a fraction, m , of individuals as immigrant in each generation from a source population would not diverge unless N m < 1, and HALDANE (1930) showed that immigration would overcome the effect of natural selection (s) in favor of an allele in a single population unless m/s < 1. These results have provided the general guidelines about the relative strength of gene flow, and they have been extended to more complicated models of population structure (e.g., MARUYAMA 1970, 1972; LATTER 1973; NAGYLAKI 1975, 1978, 1983; FELSENSTEIN 1976; fora most recent review, see SLATKIN 1985). However, in spite of analytical results showing that even a small amount of gene flow can be effective in preventing population differentiation (WRIGHT 1951), until very recently the role of gene flow in population structure has been assumed to be of less importance. This is because the early field studies in population genetics showed that gene flow in most Genetics 1 1 5 313-322 (February, 1987) species was highly restricted (DOBZHANSKY and WRIGHT 1943; BURLA et al. 1950; WALLACE 1966; RICHARDSON 1967; CRUMPACKER and WILLIAMS 1973; DOBZHANSKY and POWELL 1974; JOHNSON and HEED 1976; ENDLER 1977). And second, the polymorphic genetic systems most vigorously studied in the 1950s and 1960s, especially chromosome inversion polymorphisms (e.g., see ANDERSON et al. 1975), generally showed significant geographic differentiation in most species, which was taken to mean that gene flow was less important regardless of its magnitude in nature. Recent developments in experimental as well as theoretical population genetics have shed new light on gene flow. Recent field studies with Drosophila have shown that dispersal rate is much higher than previously thought. In a migration study using marked flies,JoNEs et al. (1 98 1) showed that dispersal between isolated populations of Drosophila pseudoobscu? a was quite extensive; marked flies could move at least 10 km in 24 hr over a desert. In a similar study, COYNE et al. (1 982) showed that in D. pseudoobscura extensive long distance migration can occur over a short period of time (15 km in 15 hr), even between favorable habitats. The latter study also included “yellow” flies, i.e., Drosophila melanogaster and Drosophila simulans, which showed similar long distance migration except 314 R. S. Singh and L. K. Rhomberg that the recapture frequencies were lower for them than for D. pseudoobscura. A second reason for renewed interest in gene flow is the availability of models of gene flow and genetic drift, and gene flow and selection that can be applied to multilocus data as available from gel-electrophoretic studies. Because genetic drift and gene flow affect all loci in the same way (CAVALLI-SFORZA 1966), estimates of migration rate and/or population size can be used to generate patterns of geographic variation for selectively neutral genes that can serve as null hypothesis. We have completed an electrophoretic survey of 15 worldwide populations of D. melanogaster for 1 17 gene-protein loci. Some of these populations are presently being studied for genetic variation of two-dimensional gel electrophoretic proteins, mitochondrial DNA and heat shock proteins. The overall goal of this series of studies is to generate a broad-based genic variation data set so that patterns of geographic differentiation and effects of functional constraints on genic variation can be studied with a single species; we need not pool data from a variety of sources or organisms which makes the analysis statistically robust but biologically uninterpretable. In this report we first provide estimates of gene flow based on the spatial distribution of private alleles (SLATKIN 1981, 1985) in different sets of continental and global populations of D. melanogaster. These estimates suggest that gene flow among geographically separated populations of D. melanogaster is quite extensive. We then show that in spite of extensive gene flo~7, West-Coast populations of D. melanogaster (comparable geographically to D. pseudoobscura populations) show significant geographic differentiation and latitudinal clines. These results, in combination with the recently reported similar long distance migrations in D. pseudoobscura and D. melanogaster (COYNE et al. 1982), suggest that migration in combination with natural selection rather than migration alone is probably responsible for the uniformity of allozyme polymorphism in geographic populations of D. pseudoobscura (LEWONTIN 1974). MATERIALS AND METHODS The eight West-Coast (North American) populations of D. meLanogaster were studied to provide data that could be compared with those from D. pseudonbsura (Figure 1). These are from Port Coquitlam, British Columbia; Carnation, Washington; Medford, Oregon; and from Hamilton, Sonoma. Fresno, El Rio, and Lakeside, all in California. Samples from Sonoma and El Rio were kindly provided by Dr. MARGARET KIDWELL (Brown University) and the rest by Dr. FRED KOHAN (University of California, Davis). Sixteen to 20 isofemale lines, sampled between 1979 and 1982, were studied from each population. A total of 20 enzyme loci were studied (Table 3). Enzyme loci showing latitudinal clines in previous studies (SINGH, HICKEY and DAVID 1982) were preferentially included in this study. Electrophoresis was done on vertical polyacrylamide slab gels. The electrophor-esis buffer and staining procedures for alcohol dehydrogenase (ADH), aldehyde oxidase (AO), esterase-6 (EST-6), esterase-C (EST-C), glucose-6-phosphate dehydrogenase (G-GPD), 6-phosphogluconate dehydrogenase (6-PGD), leucine aminopeptidase-D (LAP-D), octanol dehydrogenase (ODH), and phosphoglucomutase (PGM) are published or referred to in SINCH, HICKEY and DAVID (1982). The electrophoresis and staining procedures for carbonic anhydrase, diaphorase and hexokinase were adopted from HARRIS and HOPKINSON (I 976). Gene flow was estimated from the spatial distribution of rare alleles following the methods developed by SLATKIN (1 98 1, 1985). T h e data used in this analysis consist of allele frequency distribution of 20 protein loci in eight West-Coast populations (Table 3) and 1 17 protein loci in 15 worldwide populations of D. melanogaster (Figure 1) (SINCH, HICKEY and DAVID 1982; R. S. SINCH and L. K. RHOMBERC, unpublished data).