INDUCTION OF GENE POOL DIFFERENTIATION IN DROSOPHILA MELANOCASTER

The selective effect of the ecological parameter "food medium" upon the gene pool of Drosophtla melanogasier was investigated in nine cage populations b\ e-itimating the alloz\me frequencies of the a-Gpdh and Adh loci. The differentiation obsened was found to depend dramatically upon the action of the aboke mentioned environmental factor. Small differences may be mainl) attributed to the effect of the differing genetic backgrounds of each gene pool. When the environment was restrictive, a dramatic gene pool differentiation was observed. A discussion is provided bearing upon the observed induction of a process which involves large groups of coadapted genes upon which selection actually acts. Furthermore. the role of nutrition. and especially yeast. was found to have a profound effect upon the observed genetic differentiation of the gene pool of D. tnelanoqasier. Neui populations en cage ont ite utilisee'i Ion d'une etude sur les effets selectifs du paramitre Gcologique "nouiriture" sur Ie pool ginique de Drosophila melanogasier. La fkquence des allozymes des locus a-Gpdh et Adh aete determinee. et les differences que 1.011 a observees semblent fortement dependre du facteur pkcite. Les ligers icarts Â¥ion principalement attribues aim differences fondamentales des divers pools geniques qui pksentent des distinctions tres importantes en milieu restrictif. La discussion qui sun les resultats suppose I'induction d'un mecanisme irnpliquant d'importants groupes de genes coadaoiis sur lesquels se dinge ['action du processus de selection. Par ailleur'i, la nutrition et surtout la le\ure semblent agir profondement sur la diffirenciation genitique des pools de D . rnelans~asier. [Traduit par Ie journal]


Introduction
The study of the genetic structure of populations is one of the approaches to understanding the process of evolution. If one wishes simply to comprehend the evolutionan forces or to control them, it is helpful to evaluate the significance of the effect of certain ecological factors upon gene pool differentiation. Many studies have reported on temporal and geographical genic variations in natural populations associated with specific environmental conditions (for review see Parsons, 1973;Nei. 1975). However. data bearing upon these questions may be more appropriately obtained from laboratory' cage populations where conditions may be readily controlled.
In kiew of the above. an attempt was made to observe the reaction of a large gene pool of Drosophila melanogaster to the ecological parameter "food medium" ("poor" food mediumrestrictive environment; "rich" food mediumoptimum environment). It is well known that yeast is the primary food source of Drosophila (Sang, 1949) and that yeast-culture media are very suitable for Drosophila cultures (Demerec. 1965). Thus. food medium poor in yeast generates a restrictive environment in which the flies live under limiting (competitive) conditions, while food medium rich in yeast defines an optimum (or near optimum) environment with no or little competition (Alahiotis. 1976). This investigation examines the idea that should strong competition for food and living space exist, natural selection will occur causing gene pool differentiation. In order to trace such a possible gene pool-environmental reaction. the allozvme frequencies of a-Gpdh (a-Glycerophosphate dehydrogenase) and Adh (alcohol dehydrogenase) were utilised. These two enzyme loci were found to associate nonrandomlk with the polymorphic inversions of the second chromosome in the populations examined here ( Alahiotis et a l . , 1976). Hence, they can be considered as markers of large blocks of genes.

Material and Methods
In our experiments. nine cage populations were studied for the allozyme frequency changes of two enzyme loci, over 30 generations. under the influence of the ecological factor food medium. Populations 1A. IB:,, 18,. and 2A were raised at 2520.5'C. 43Â±4[7 RH (relative humidity) in a poor corn meal-agar-sugar food medium ( 12.5 g corn meal, 1.8 g sugar. 1 g agar per 100 HzO, with 1 rnl of diluted living yeast suspension per food vial) to which propionic acid was added. Populations. 1 9 , IB,. lB3, 2B, l A, were exposed to the same conditions of temperature and relative humidity but raised in a rich dead-yeast-agar-sugar food medium (10 g dead-yeast. 10 g sugar. 3 g agar per 100 ml HzO) where again propionic acid was added. Care was taken to have more o r less the same amount of medium in all the vials used for the experiments. All these populations were maintained under a 12 h darklight cycle. During the early generations there was a rapid expansion of population size, which subsequently was regulated according to the kind of food medium on which each population was raised. Thus. the rich food medium populations had produced approximately the same number of flies fabout 2.800) while. the poor-food medium population size was about 1.100 flies. (For more details see Alahiotis. 1976.) Populations IBi. IB, as well as IB,. IBi. were derived from the basic population 1B ten generations after the latter originated. The previously mentioned populations were put two-by-two ( 19,l Ba, 1 B3-18,) under the effect of the food medium (rich-poor food medium). Populations IA, 19 originated from a common gene pool of 600 parents collected from a Cephalonian-Greek natural population. while populations 2A. 2 9 originated from a Gavros-Achaia-Greek natural population. The above mentioned 600 parents originated from 100 isofemale lines derived from recently collected (summer 1973) individuals fpopulations were set up 20-30 days after the capture of flies). Six ( 3 virgin females and 3 males) 4-6 days old individuals were taken from each of the isofemale lines and thus the 300 virgin females and the 300 males gathered were allowed to mate randomly in the cage (1B or 2B). Populations 1A and 2A were derived by replicating the 1B or 2 9 populations respectively. Consequently. we ma\ regard the derived population-) of each series. as possessing practically the same gene pool. Moreover. in order to provide more evidence, a complementary experiment was carried out involving a new cage population (the IA,) which was produced from the 16th generation of the 1 A population.
The populations were maintained in plexiglass cages the dimensions of which were 41 X41 X 16 cm. possessing 14 vials each (the dimensions of which were 1 O X 2.5 cm). For details see Stem ( 1972).
In sampling the cases, three or four fresh food vials were placed in the cage. After 24 h the vials were removed and care was taken to avoid crowding in them. The vials were then placed under the same environmental conditions in which the respective populations were originally maintained. and emerging adults were scored for a -G p d h (E.C. 1.1.1.8.) and Adh loci fE.C. 1.1.1.1 .) genotjpes. In order to characterize the enz>me genotypes of individual flies. starch gel electrophoresis was carried out at 300 volts. 100mA. for about one hour. using 0 . 5 M Tris versene-borate buffer pH 8.0 fShaw and Prasad. 1970). Cells were sliced horizontally: the lower half was stained according to the method of Shaw and Prasad (1970) for a-Gpdh. while the upper half was stained for Adh using the method of Johnson and Denniston ( 1964).

Results
Our findings are presented in four tables and four figures. In all cage populations studied. two allozyrnes for each locus were found. Initial alloz>me frequencies used were the frequencies found in the two aforementioned natural Southern Greek populations.
As shown in Table I. the allozyrne frequencies of the a-Gpdh and Adh loci remain the same over 30 generations for the rich food medium populations ( 1 B . I Bl. 1 Bz. 2 B ) with a slight difference as regards to the AdhF frequency in population IBl which wa.-> contaminated with mites for some generations.
In contrast. a dramatic change was observed in the allozyme frequencies of the a-Gpdh and Adh loci for the poor medium populations over 30 generations (Table 11). The fast allozyme of the a-Gpdh locus revealed an increase in its frequency for all populations studied while the Adh F, allozyme exhibited a decrease in its frequency for all populations studied. It seems that strong directional selection was operating throughout the duration of the experiment. In some cases (population 2A) an elimination of one allele occurs. Figures 1-4 illustrate in a graphic way those of our findings which we considered to be of great interest. Dramatic changes are observed in allozyme frequencies when the populations are under competitive conditions (poor food medium population), whereas these frequencies remain stable under rich food conditions. In addition, good evidence for the effect of the "food medium" parameter upon allozyme frequency differentiation is demonstrated by cage population 1 A,. In this case we observed (Table 111; Figs 1, 2) a dramatic reversal. in the direction of allozyme frequencies for both loci when the population is treated first in the poor food medium (population 1A) and then in the rich one (population lA,). After a few generations the new frequencies tend to approach  those of populations maintained in the rich food medium ( 1 B, 1 Bi. 1 Bz, 2B). As far as the Adhf frequency differentiation in the 1A population is concerned, it appears (Fig. 2) that there is a parallel increase of this frequency in both the 1A (poor medium) and lA, (rich medium) populations, mainly between the period 20-25 generations. However, this is not true in later generations. Fig. 2 demonstrates an oscillation of the AdhF frequency in the 1A population with a general decrease over the entire 30 generations resulting in a clear frequency differentiation between these populations. The difference in the AdhF frequency between 1A and lAi at the 30th generation is highly significant (xf=9.6658 Pc.01; according to Workman and Niswander 1970). It is possible that sampling errors or random drift may have been the contributing factors for the oscillation observed (unexpectedly high frequency of AdhF in 20-25 generations) in this population. Moreover, Table I1 shows clearly that, as a general rule, the AdhF frequency decreases in every poor medium population throughout the 30 generations examined.
The results of testing for linkage equilibrium between the two loci (by comparing the observed with the expected combinations of the genotypes of a-Gpdh and Adh; Li, 1961) are given in Table IV. and show that these two loci are not generally in linkage equilibrium. Furthermore, in a recent-publication (Alahiotis et at.. 1975) we have shown that nonrandom association exists between the allozymes of these loci and the ( G e n e r a t i o n s  In(2L)22D-34A and ln(2R)52A-56F inversions, both in cage populations as well as in a natural one. Thus, the existence of linkage disequilibrium between the alloz\mes ofAdh and a-Gpdh loci may be considered the result of the presence of these inversions in the population studied. On the other hand, the fact that these two loci belong to two different metabolizing systems (Kojima et a / . , 1970), decreases the probability of gene interaction between them.

Discussion
In the present investigation an attempt was made to induce gene pool differentiation under the influence of the environmental factor "poor-rich food A test for linkage equilibrium between the a-Gpdh and Adh loci in cage populations of Drosophila melanogaster

Populations
Probability Sample Populations Probability Sample (generations) ( P) size (generations) (P) size medium". From our findings, it appears that the aforementioned differentiation is dramatic. The observed change in the allozyme frequency of a locus is in a specific direction for all populations studied under similar environmental conditions, thus suggesting the operation of directional selection. Furthermore. by utilizing the allozyme frequencies found in (Greek) natural populations as initial frequencies of the allozymes for the cage populations. we may assume that a given deviation from the initial (equilibrium) frequencies is due to the parameter under study. Another factor worthy of consideration and capable of causing differentiation in the variability is genetic drift. In our experiments however, one could argue against genetic drift having played an important role on the basis of the follow ing reasons: 1 ) large population sizes such that consecutive generations overlapped each other extensively. 2) the findings that changes in gene frequencies under the same environmental conditions were similar in many populations (small differences between populations with different original gene pool may be attributed to different genetic backgrounds), 3) the detection of a reverse frequency differentiation in the l A i population. 4) the existence of a proportional frequency differentiation in lethal bearing chromosomes (Alahiotis. 1976), and 5 ) in most of the cases our samples for each test were rather large. One cannot rule out the possible effect of genetic drift by the above arguments for the population sizes in poor food medium population were small (N= 1.100). However. the maintainance of high lethal bearing chromosome frequencies (0.3900-0.4041; Alahiotis, 1976) in the same populations suggests the existence of a large S e (effective populations size). The same is true. at least. in one of the natural populations ( Yannopoulos. 1974) from which the founders were captured.
The density achieved by the cage populations studied depends on the type of food medium (poor food medium. high density-rich food medium. low density) used. Competition is one means by which selection can take place. Thus. competition in poor food medium populations must be high, while. it is low in the rich food medium populations. Here. one can onl! speculate that "mutual facilitation'' may be of importance in polymorphic situations (like ours) here the different genotypes segregating in a population could have slightly different ecological requirements. thus resulting in a mutual facilitation (see Parson, 1973. for review ) . However. more data (e.g. viability experiments) are needed so as to assess whether or not the above situation is true for our populations. It must also be noted that the small amount of live yeast suspension used in the poor medium is consumed by the flies in a few hours so that there is no live yeast left during larval development; hence, we can assume that competition exists for food (yeast) mainly in the larva stage. Thus, the observed dramatic gene pool differentiation could be attributed to the competitive conditions which reflect the difference of the food media. From this point of view it appears that nutrition, and especially yeast. may have some serious effect on gene pool differentiation and ultimately on the evolution of Drosophila melanogaster. It is important to note that we found strong sexual isolation (Isolation Index e0.35) in later generations among the same populations.

Generations
As already noted, the allozymes at both the a-Gpdh and Adh loci are nonrandomly associated with inversions on the second chromosome. The Adhs and a-GpdhF alleles are in strong linkage with the In(2L)22D-34A, thus suggesting that the frequency of the ln(2L)22D-34A is much higher under competitive conditions than its' standard (see also Alahiotis et al., 1976). These associations do not allow us to attribute the changes in allozyme frequencies to a selective effect at the loci studied. They may very well reflect a process which involves large groups of coadapted genes upon which selection acts Lewontin, 1968, 1971). The proportional frequency differentiation of the lethal bearing chromosomes (Alahiotis, 1976) as well as of the gene arrangements (Alahiotis et a l . , 1975) and the linkage disequilibrium found between inversions of the second and third chromosomes (Alahiotis et al., 1976) or between the alleles of the two loci studied, favors the above view.
However, in recent studies it has been suggested that allozymes do not constitute part of the coadapted complexes of genes characterizing inversions (Zouros, 1976). Moreover, linkage disequilibrium between allozymes and inversions is considered to be a historical accident where allozymes share their fate with the inversions (Watanabe and Watanabe, 1977). However, by studying additional cage populations (Alahiotis and Pelecanos, 1977) it has been shown that the a-Gpdh allele frequencies were found to be uniform in relation to different second chromosome inversion frequencies { I n ( 2~) 2 2~-3 4~} , a fact which supports the view that while this locus is included in the inversions. it doesn't share its fate (e.g. in cage populations IB. ID, natural Greek population (f) a -G p d h F = 0.61 36, 0.6293, 0.6769 and (f) ln(2L)22D-34A = 0.061 1. 0.4375. 0.0462 respectively). Furthermore, with respect ot the Adh locus, there is similar Adh allele frequency-temperature interaction in populations with low (Natural populations : Pipkin et a1 ., 1976) and high (our cage populations for the temperature effect: Alahiotis and Pelecanos. 1977) second chromosome ( 2 L ) inversion frequencies Finally, since gene frequencies tend to be similar under the same environmental conditions in all cage populations studied, this favors the coadaptation hypothesis.