Latitudinal clines of allelic frequencies in Mediterranean populations of Ceratitis capitata (Wiedemann)

Clines latitudinaux de frequences alleliques dans des populations mediterraneennes de Ceratitis capitata (Wiedemann). Des echantillons de la mouche des fruits, Ceratitis capitata, provenant de 6 regions differentes de 4 pays mediterraneens, ont ete examines pour leur polymorphisme genetique a 25 locus enzymatiques. Six des locus se sont averes polymorphes et les frequences genotypiques observees a 4 de ces locus sont conformes a l'hypothese de panmixie. Un cline latitudinal significatif des frequences alleliques a ete mis en evidence pour 2 de ces genes polymorphes par un test de correlation de rang de Spearman. Pour 4 locus, les variations de frequences alleliques dans l'axe nord-sud sont importantes, par exemple la frequence de l'allele 1.00 du gene Idh varie de 1,00 a 0,63 le long de cet axe.


INTRODUCTION
The Mediterranean fruit fly Ceratitis capitata is a polyphagous and multivoltine tropical and subtropical species, and one of the most serious pests of fruits and vegetables. During the last 150 years, medfly has been established in several countries including those of the Mediterranean basin from its proposed equatorial African origin (Fletcher, 1989). Even though the first report of this pest within the European Mediterranean area dates from the middle of the last century, the medfly has expanded throughout this basin because this region offers an increasing number of host fruits of different cultivars and/or species (De Breme, 1842;Martelli, 1910;Fimiani, 1989). Medfly is a serious threat to important fruit production centers throughout the world. However, the biology of this species is poorly known, especially its population genetics. Genetic variability might also provide information on the spread of C capitata (Kourti et al, 1990;Gasperi et al, 1991). Recent results showed large genetic differences between introduced populations and those postulated to be their ancestral African ones. The average heterozygosity of medfly populations within the Mediterranean basin is about 5% versus 22% in the African populations (Huettel et al, 1980;Kourti et al, 1990;Gasperi et al, 1991). This significant heterozygosity of African populations (22%) is comparable to that of other insect natural populations, eg, Drosophila (Ayala et al, 1972;Lewontin, 1974;Prakash, 1977;Hyytia et al, 1985). The similarity displayed in allele frequencies among medfly or Drosophila populations could be expected under the neutral theory, provided that some exchange of genetic material takes place between populations (Kimura and Marayama, 1971; Kimura and Ohta, 1971). In many cases, it was concluded that the subcosmopolitan status of medfly was recently achieved, due to human transport, and that any genetic divergence between geographically distant populations has occurred over a very short time. However, C capitata appeared to be adapted to different environmental conditions, being able to constitute large populations in both the tropics and the temperate regions. Since the polymorphism of the introduced populations is low compared with that of their African counterparts, we focus our effort on elucidating this discrepancy among populations originating from different Mediterranean areas. This study was designed to detect latitudinal clines in gene frequencies. For this purpose, 6 natural populations from distantly located areas were examined.

Origin of populations
Six populations of C capitata from the Mediterranean basin were used in this study. The Spanish population was represented by a sample of pupae on peaches collected in October 1986 in the area of Castellon (S). The Greek populations were: a) from Attiki, where pupae were collected on peaches in July of 1986 (G1); and b) from the island of Crete, where pupae were also collected on peaches in July 1986 in the area of Chania (G2). The Israelian population was from the area of Best Dagan, where pupae were collected on apricots in June 1984 (I). The Egyptian populations were represented by 2 samples: a) pupae on apricots collected in May 1986 in the area of Kalubia (E1); and b) pupae on apricots collected the same period in the area of El Fayum (E2). Collections of wild flies from these Mediterranean populations were made by harvesting infested fruits from the ground. An effort was made to collect samples from 1 food source: the stone fruits (peaches or apricots).

Statistics
Chi-square tests were performed to compare observed numbers with those expected under Hardy-Weinberg equilibrium. The Chi-square test of the correlation between genes (Barker et al, 1986) was applied for loci with more than 2 alleles. The degree of relationship between clinal patterns was determined by Spearman rank correlation (Statgraph Statistical Program). The strategy is to look at conditional frequencies and their correlations with latitude as clinal measures.

RESULTS
The allelic frequencies at the 6 polymorphic loci, the collection sites and the number of individuals analysed are listed in table I. Most loci have 2 alleles with the exception of G-6-pd. All the peptidases are unmapped while the position of the other Alleles in order of increasing anodal mobilities; n, number of individuals analysed. Populations according to country of origin. The numbers in the row below the countries represent latitudinal positions.
At the 2 loci, G-6-pd and Idh showing significant correlations with latitude, the common allele frequencies exhibit a gradual change from different localities of collections, indicating the presence of a latitudinal cline (fig 1). In all cases, there is a large frequency difference between the third site of collection (Crete) and fourth (Israel) (fig 1). The same difference in G-6-pd is smaller. In contrast, the frequency changes of the most common allele at the locus Mpi are rather smooth with the exception of the population collected from Spain   (fig la). For loci showing the smallest frequency difference, Pep-1 and Pep-2, these are 1.1 and 4.3%, respectively. The differences of loci that showed latitudinal clines, G-6-pd, Mpi, Pep-3 and Idh, are 19.9, 21.3, 27.3 and 36.7% respectively. The Mpi frequency difference between natural populations collected from Egypt and from Attiki is 33% and is one of the highest observed.

DISCUSSION
An extensive analysis has shown that the polymorphism (at least 16 genes out of 25) found in the African populations of C capitata was high. However, only 6 loci out of these 25 were found polymorphic in the introduced populations of Mediterranean medfly (Kourti et al, 1990). It is interesting to note that these 6 loci, Mpi, G-6-pd, Idh, Pep-1, Pep-2 and Pep-3 maintain variants at substantial frequencies and from these, the frequencies of Mpi, G-6-pd, Idh and Pep-3 change steadily with latitude. When neighboring populations of a species are compared, one finds that they usually differ from one another, slightly or appreciably, in a number of characteristics, eg, size, color or any other morphological or physiological character (Halkka et al, 1975;Rhomberg and Singh, 1989). Huxley (1942) introduced the term 'cline' for such a character gradient. The gradual change of the common allele frequencies for at least these 4 loci corresponds to the latitudinal cline definition. However, only 2 loci, G-6-pd and Idh, have shown a significant r s . The cyclic behavior of Mpi, found by Malacrida et al (1992)  Adh, a-Gpd, G-6-pd, Odh, 6-Pgd, Aph, Est-C, and Mdh (Voelker et al, 1977(Voelker et al, , 1978Anderson, 1981;Oakeshott et al, 1981Oakeshott et al, , 1982David, 1982;Anderson and Oakeshott, 1984). The seasonal cycle in Mpi and the latitudinal clines of other loci suggest that alternative alleles have different optimal temperatures or other environmental variables (see also Tomiuk and Wohrmann, 1984). The working hypothesis is that the cline results from a geographically variable selective factor of the environment (Halkka et al, 1975). A concordance of clines for different characters is normally only found where ranges are essentially latitudinal and where the various environmental gradients (temperature and humidity for example) run more or less in a parallel way. The study of geographic variations has revealed that many of them are clinal. Clines are, ultimately, the product of 2 conflicting forces, selection, which would make every population uniquely adapted to its local environment, and gene flow, which would tend to homogenize all populations.
While some alleles sampled at high frequency in African populations are also found in Mediterranean populations, others, especially the uncommon ones, are locally or regionally lost (Kourti et al, 1990). This new pattern of allozyme frequency is maintained at a relatively stable level.
The contrast in polymorphism patterns between African and Mediterranean populations suggests that medfly polymorphism could be maintained by balancing selection for which the equilibrium shifts with climate. Therefore the adaptively neutral or nearly neutral variation is expected to be purged, most likely by the repeated passage of local populations through bottlenecks (Kourti et al, 1990).
Moreover, the lower polymorphism of Mediterranean populations compared with the African populations resembles the general pattern of marginal populations of a species submitted to genetic drift. Loci that are under rather strong balancing selection manage to maintain their total variability while others become monomorphic. Identification of such loci, however small in proportion, as in the case of medfly, may be more important than performing generalized tests of the neutral theory (Tomiuk, 1987). This explanation of widespread monomorphism, the difficulty of establishing and maintaining alleles expect by balancing selection, is an alternative to the hypothesis of broad adaptability of 'general purpose genotypes' (Parker et al, 1977;Angus and Schutz, 1979;Jaenike et al, 1980;Lynch, 1983). Environment (latitude and/or season) could exert its force on allozyme frequencies. Since it is reasonable to expect that some loci might be important selective targets, and that there is a relationship between genotype and environment (often confirmed), as previously discussed, neutrality could not satisfactorily explain the data. Similar results have been reported for other natural populations (Johnson et al, 1969;Johnson and Schaffer, 1973;Voelker et al, 1978). In most cases there are correlations between allelic frequencies and environmental variables clearly favoring the selection hypothesis (Schaffer and Johnson, 1974).
In conclusion, the genetic latitudinal variations of C capitata could be considered, as a whole, as a consequence of natural selection. Further studies, comparing tropical and temperate populations from other parts of the world, will determine whether such clines exist worldwide.

ACKNOWLEDGMENTS
We would like to thank K Krimbas for critical reading and attentive comments of this manuscript. AK is thankful to M Loukas for introducing her to the C capitata system.