Proc. Nati. Acad. Sci. USA Vol. 82, pp. 8552-8556, December 1985 Evolution Detection of simple polygenic segregations in a natural population (genetic variation/Drosophila melanogaster/dominance modifiers) JAMES N. TH6MPSON, JR.,* AND C. G. N. MASCIE-TAYLORt *Department of Zoology, University of Oklahoma, Norman, OK 73019; and tDepartment of Physical Anthropology, Cambridge University, Cambridge CB2 3DZ England Communicated by Hampton L. Carson, August 7, 1985 ABSTRACT Penetrance frequencies were used to quantify MATERIALS AND METHODS segregating polygenic effects in a natural population of Drosophila melanogaster. When males from a series of 100 Polygenic modifiers offormation ofthe fifth-longitudinal (L5) isofemale strains were crossed to females from a veinlet line vein were scored in 100 isofemale strains collected around a (ve) single watermelon on a single day in 1973 in Oklahoma City, that had been selected for shortened veins, gaps commonly Oklahoma. The bait was exposed for less than 6 hr before appeared In the fifth longitudinal (L5) vein in the ve/+ collections were made. A localized natural D. melanogaster heterozygotes. We were able to assign each strain to one of six population is extremely difficult to identify with certainty. In significantly different clusters, based upon the pattern of sampling 100 separate genomes collected at the same time polygenic modifiers of ve dominance segregating in each strain. from the same location, however, we have done our best to We conclude that the extensive range of phenotypic variation obtain flies from the same population. in vein-forming ability is actually based upon a relatively Each female had been inseminated before collection, simple underlying polygenic structure that is consistent with though the number ofinseminations could not be determined. segregation of only a small dumber of alleles or allele combi- Flies were immediately isolated in separate food vials, and nations in the wild population. the progeny of each female were mass-transfei-red at each generation to minimize sampling effects in culture. Each Polygenic segregation plays a central role in the adaptability isofemale strain consequently carried a limited sample of the of a gene pool. Yet only a few studies have attempted more total allelic variation segregating in the population. than a superficial analysis of the polygenic structure of The manipulation of environmental variables or the pres- natural populations (1-5). This is not particularly surprising, ence ofa major mutation has frequently been used to uncover cryptic polygenic variation (10). Such variation can be traced since polygenic loci are rarely identifiable individually and to the segregation of polygenic alleles that have effects not their effects are often masked by environmental factors. In usually expressed overtly in typical wild-type individuals. In spite of these limitations, however, some aspects of this study, the polygenic modifiers of vein formation in each polygenic architecture are detectable under highly controlled isofemale strain were categorized by crossing 10 single males conditions, as biometrical analysis (6), whole-chromosome from each line to females from a veinlet (ve, chromosome 3, substitutions (7, 8), and isofemale-strain surveys (9) clearly locus 0.2) line that had been selected for reduced vein length demonstrate. (ve Short). This selected veinlet line provided increased One key conclusion from such studies is that relatively few sensitivity in the detection of segregating modifiers, partic- loci are required to account for the majority of the response ularly those affecting the L5 vein (12, 13). to artificial selection (10). Unfortunately, the degree to which The frequency ofterminal L5-vein gaps (i.e., gaps between this can be extended to natural populations is almost com- the end of the L5 vein and the wing margin) was determined pletely unknown. Although many loci could theoretically from 20 male and 20 female wings scored per culture (a total contribute to quantitative variation in a population, a limited of400 wings from each ofthe 100 isofemale strains). Only one number of polygenic combinations may occur at high fre- wing was scored from each fly, and gaps in other veins were quency as a result of selection for polygenic balance. Indeed, rare. Single males had been used as parents of each culture. this is a major, testable prediction from Wright's shifting- Thus, the 40 wings scored for each replicate provide an balance theory of genetic architecture (11). estimate ofthe average ve/+ gap penetrance associated with In this study, we outline a method by which polygenic the samples of chromosomes from the segregating isofemale can be assessed in natural and we line. segregation populations, Cluster analyses were carried out by the unweighted pair test the prediction that only a limited number of polygenic group method with arithmetic averages (UPGMA; ref. 14) combinations are common in a population. To do this, we using NT-SYS, a package ofmultivariate statistical programs have surveyed polygenic modifiers ofwing-vein formation in (15). Other statistical analyses were performed using the 100 separate isofemale strains of Drosophila melanogaster statistical package for the social sciences (SPSS; ref. 16). collected from a single location. The polygenic makeup and All lines and crosses were maintained on an agar, oatmeal, patterns of segregation in these lines were studied by using and molasses medium seeded with live yeast. Temperature cluster- and discriminant-analysis techniques. The results was 25 ± 0.50C. strongly suggest that, at least at one level, the polygenic structure of a natural population can be comparatively simple. RESULTS The publication costs of this article were defrayed in part by page charge The Oklahoma population is highly polymorphic for payment. This article must therefore be hereby marked "advertisement" polygenic modifiers of vein formation. This is readily seen in in accordance with 18 U.S.C. §1734 solely to indicate this fact. Fig. 1, which shows mean L5-vein-gap frequencies in 100 8552 Downloaded by guest on September 29, 2021 Evolution: Thompson and Mascie-Taylor Proc. Natl. Acad. Sci. USA 82 (1985) 8553 30 - 20- E 10 z 0 50 100 150 200 250 Total number of wings with gaps FIG. 1. Distribution of mean frequencies of L5-vein-gap expression in 100 isofemale strains of D. melanogaster. Each mean is based upon average ve/+ penetrance in 10 replicate crosses. genetically heterogeneous isofemale strains. The phenotypic loci would increase the complexity of segregational patterns, variation among crosses is high, ranging from 0 to 40 vein as would environmental factors and differences in allele gaps in the 40 wings scored per cross. Superficially, this frequency within an isofemale line. approximates a Poisson distribution, although there is evi- Thus, this experimental approach is based upon a simple dence of significant residual variance (Kolmogorov-Smirnov prediction. If segregational patterns characteristic of each test, P < 0.001). isofemale line fall into a small number of significantly Variation in gap penetrance is not due to genetic heteroge- different categories, the polygenic basis of the trait must be neity in the ve Short selection line used as the outcross standard. comparatively simple. There are several statistical ways to This was shown by mating each of 75 single males from 75 determine whether an underlying group structure exists different isofemale lines to 2 virgin ve Short females. The within a data set. For this study, we have used cluster- females were then separated and placed singly in culture vials. analysis techniques in the NT-SYS package to evaluate the A total of 20 male and 20 female wings were scored in each of overall degree of similarity in the distribution of gap frequen- the replicates of each tested male. Although there was highly cies in comparisons among all pairs of isofemale strains. significant variation among males (X74 = 1020, P < 0.001), The function of cluster analysis is to assess the degrees of variation among ve Short females (i.e., variation among repli- similarity among pairs of quantitative comparisons. In this cates, within males) was not significant (X25 = 79) instance, the comparisons we want to make are between pairs The correlation between males and females was 0.863. of phenotypic distributions, such as those in Fig. 2. Resem- Thus, approximately 74.5% (r2 = 0.8632) of the variance in blance is quantified by the Z value generated by the gap frequencies is due to genetic differences among lines. Kolmogorov-Smirnov two-sample test (19, 20), which is used These are almost exclusively autosomal polygenes, since to evaluate the probability that two independent samples effective sex-linked loci are rare in this sample (17). The have been drawn from the same population or from popula- remaining 25.5% of the variance can be accounted for by tions having the same distribution. It is sensitive to distribu- uncontrolled environmental effects and by sex-limited differ- tion qualities such as location, dispersion, and skewness (19). ences in development (for example, males commonly express In our study, the Kolmogorov-Smirnov Z values are used vein defects more strongly than do females; ref. 13). as measures of "distance"-that is, as measures of the A phenotypic distribution such as this is consistent with degree of similarity or separation between pairs of L5-vein- most models of the polygenic structure of natural popula- gap phenotype distributions. Hierarchical cluster analyses tions. A continuous distribution is commonly considered were performed to identify isofemale strains that produce strong evidence for a broad range of underlying genotypes similar phenotypic distributions of test progeny. The produced by segregation at many loci. However, the infor- dendrogram in Fig. 3 illustrates the six distinct clusters of mation that is actually provided by such analyses is also strains that are revealed by this analysis. compatible with the idea that only a few genes or gene Separate cluster analyses taking into account the slight complexes are segregating (18).