Myosin-heavy-chain DNA polymorphisms of subterranean mole rats of the Spalax ehrenbergi superspecies in Israel

Restriction-fragment-length polymorphisms (RFLPs) of the sarcomericmyosin heavy chain multigene family were studied in 13 populations of 4 chromosomal species (2n = 52, 54, 58 and 60) of the mole rat Spalax ehrenbergi superspecies in Israel. A minimum number of 6 hybridizing fragments occurred in the mole rat, corresponding to a minimum of 6, probably even 8, different loci of the sarcomeric myosin heavy chain. The level of polymorphism was 31%, with a low number of alleles in the polymorphic loci. All of the populations, including a desert isolate, were polymorphic at some level. Amongsample fragment variation was adequate to differentiate between some of the chromosomal species, and confirms earlier evidence that gene flow between the chromosomal species is limited.

The adaptive radiation of the S ehrenbergi complex is closely associated with aridity gradients. Hence, this adaptive radiation is associated with distinct climatic diversity: 2n = 52 radiated in the cool-humid upper Galilee mountains; 2n = 54 in the cool -semi-dry Golan heights; 2n = 58 in the warm -humid lower Galilee mountains, central Yizrael valley and coastal plains; and 2n = 60 in the warm -dry mountains of Samaria, Judea, northern Negev and the southern part of the Jordan valley and coastal plain (fig 1). Surveys of allozyme polymorphisms in the S ehrenbergi superspecies, as well as in other subterranean mammals (Nevo et al, 1984) revealed low levels of allozyme polymorphisms. This was explained as an adaptive strategy to the relatively constant microclimate underground. In general, DNA polymorphisms display far more variation, as compared with allozyme polymorphisms of metabolically vital enzymes.
Analyses of restriction fragment length polymorphisms (RFLPs) in Spalax ehrenbergi reveal, as expected, a higher level of polymorphism. Diversity was found in the non-transcribed spacer of ribosomal DNA (Suzuki et al, 1987) and in mitochondrial DNA . A very high level of RFLPs and a high number of per locus alleles per loci were found in the major histocompatibility complex (Nizetic et al, 1985;Ben-Shlomo et al, 1988) and in the Period-homologous sequence (Ben-Shlomo et al, 1993). On the other hand, a low level of polymorphism was found in the haptoglobin gene ). In the present study, we examined the RFLPs of the sarcomeric myosin heavy chain (MHC) gene family.
Myosin is the major structural component of the contractile apparatus of the muscle. The myosins are proteins that interact with actin to convert chemical energy into mechanical work. A myosin molecule consists of 2 heavy chains (! 200 kDa) and 2 pairs of light chains (for review on myosin structure, aggregation properties and its role in force generation, see Harrington and Rodgers 1984). Myosin developed very early in the evolution of eukaryotic organisms (Clarke and Spudich, 1977;Warrick et al, 1986).
Myosin has been classified into 3 major categories according to their abundance in different tissues and cellular compartments: sarcomeric muscle, smooth muscle and non-muscle myosins. In vertebrates, the different sarcomeric MHC isoforms are encoded by a multigene family of closely related members (Nguyen et al, 1982;Strehler et al, 1986;Mahdavi et al, 1987). There is no cross-hybridization between sarcomeric MHC and smooth muscles and non-muscle MHC (Nguyen et al, 1982;Leinwand et al, 1983). Each MHC gene displays a pattern of expression that is tissue-and developmental-stage specific, both in cardiac and skeletal muscles (Mahdavi et al, 1987). Different numbers of MHC genes have been found in different organisms. Extensive hybridization of the rat sarcomeric MHC gene probe was observed in DNA of all metazoan organisms tested from nematode to man (Nguyen et al, 1982).
The simplest pattern of only one MHC sequence was observed in 2 species of sea urchins: Strongylocentrotus purpuratus and Lytechinus pictus. The most complex pattern was found in the goldfish Carassius auratus and in the chicken, with up to 30 hybridizing fragments (Nguyen et al, 1982). Mammalian species (human, mouse, chinese hamster, rat and rabbit) exhibit an intermediate complexity with 7 -13 hybridizing bands (Nguyen et al, 1982;Maeda et al, 1987). The gene in the rat is comprised of 24 x 10 3 bases of DNA and is split into 41 exons (Strehler et al, 1986). In the rabbit, the gene is even longer, ie x5 25 kb (Friedman et al, 1984).
No populational analyses have yet been conducted on MHC polymorphism in wild mammals. A low level of RFLP, located in the flanking region, was found in human skeletal MHC (Leinwand et al, 1983;Schwartz et al, 1986). Digestion with MspI endonuclease (4-base recognition site) yielded 3 different alleles. Some allelic polymorphism of MHC genes was found in the rat (Nguyen et al, 1982). Different strains of laboratory rats exhibited distinct hybridization patterns both in cardiac-and skeletal-specific sequences. Weydert et al (1985) found RFLPs that differentiated between 2 mouse species; Mus musculus and Mus spretus. The markers were found in skeletal MHC (embryonic, perinatal and adult) as well as in cardiac MHC genes. Spalax ehrenbergi superspecies may be an excellent model for examining patterns of genetic variation in mammals. The objectives of the present research are to estimate the number of sarcomeric myosin heavy chain genes in the mole rat and to estimate the extent of RFLP divergence within and between the 4 chromosomal species.

Sampling
We examined DNA polymorphisms of the myosin heavy chain genes in 121 mole rats sampled from 13 populations representing all 4 chromosomal species of the Spalax ehrenbergi superspecies in Israel (localities are plotted in figure 1, and sample sizes and other relevant data are summarized in table I). Live animals were caught in their underground runways, brought to the laboratory and dissected. Tissues were instantly frozen in liquid nitrogen and then stored at -80°C.

DNA extraction, digestion and blotting
High molecular weight genomic DNA was extracted from the kidneys, following Holland (1983). Three different 6-base recognition endonucleases (BstEII, KpnI, BamHI) and one 4-base recognition enzyme (TaqI) were used for digestion. Twentyfive pg of genomic DNA were incubated overnight with 150 units of a given endonuclease, as recommended by the supplier (BioLabs, New England, USA). Fully digested DNA was precipitated in ethanol, followed by electrophoresis, through both 0.6% and 1.0% agarose gels for 16 h at 30 v. Thus a single sample contained about 10 pg DNA. DNA was denatured and transferred to nylon (Hybond-N, Amersham, UK) filters following Southern (1975).
The specific probe used -pA81 (L Garfinkel, Bio-technology General, Rehovot, Israel; personal communication) was a specific skeletal myosin heavy chain of a rat adult, G-C tailed into the PstI site of pBR322. The cDNA library was prepared from RNA which was extracted from hind leg muscle of adult rats (Garfinkel et al, 1982). The clone hybridized to mRNA from adult skeletal muscle only. Since the insert included several Pst I sites, restriction of plasmid resulted in some fragments. The 2 longer fragments detected the same Spalax DNA fragments. We used mainly the longest fragment of ! 0.8 kb as a probe to detect the Spala! myosin heavy chain genes. The exact location of the probe within the myosin gene in unknown.
The DNA fragment was radiolabelled by the random priming method of Feinberg and Voglestein (1984) (Amersham: 'Multiprime DNA labelling systems') overnight at room temperature. Hybridization of 32P-labelled probes with the DNA on filters was performed in 50% formamide solution (consisting of 5 x SSPE, 5x Denhardt's solution, 0.5% SDS and 10% dextran sulphate) overnight at 42°C, following the filter manufacturer's (Amersham, UK) recommendations. Filters were washed at stringency conditions of 2 x SSPE and 0.1% SDS at 63°C for 30 min. We then autoradiographed 24-48 h at -70°C, using 2 intensifying screens (Kodak rapid or super rapid).

Data analysis
The data were analyzed as allele fragment or fragment frequencies, by scoring the bands directly from the autoradiographs. We recorded a minimum number of 6 hybridizing fragments. These fragments were obtained from hybridization of both the larger PstI fragments of the probe, suggesting the existence of at least 6 different loci. Some of these loci were monomorphic, and the others polymorphic. Two fragments were considered as allelic fragments if they appeared either each alone, or both together. However, there was no case in which none appeared, eg codominant system. The identification was relatively simple, since for each restriction enzyme there were only a few polymorphic loci (tables II, III). Identification of homozygous and heterozygous genotypes of the polymorphic loci was made directly from the autoradiographs. Observed heterozygosity (Ho) and expected heterozygosity (He) (Nei, 1973) were calculated as the proportion of heterozygous individuals, per population and per locus (per restriction enzyme). Expected heterozygosity was computed from the observed frequencies of codominant alleles under the assumption that the population was in Hardy-Weinberg equilibrium.
In some cases we could not identify or find the alternative allele(s). It seems that a portion of the fragments was undetectable by the cloned probe used in this experiment. We define the portion that could not be detected as 'noneallele' (similar to the null alleles of isozymes). In these cases we analyzed the presence of the allele and its absence (none-allele). Therefore the frequency of the present fragment might be an overestimate of the allele frequency, since possible heterozygotes were scored as homozygotes. Consequently, the none-allele frequency might be an underestimate.
We assume that most, if not all, of the mutations which caused polymorphisms are located within the myosin heavy chain genes. The gene in all mammals tested is very long, comprising > 20 kb. There is no reason to suggest smaller genes in the mole rats. Most of the polymorphic fragments we recorded were small (< 10 kb and in many cases, even < 5 kbtable II). Thus the probability that the polymorphism was within the 20 kb of the gene was very high. However, we could not specify whether a mutation was in the exons or in the introns. We suggest that all the mutations we scored were located outside the region of the probe since the 2 probes tested showed the same pattern of bands, and we found either only 1 different allelic band, or no band in the polymorphic fragment. If the mutation occurred within the probe region, each change in a restriction site should have affected the 2 corresponding fragments. Further study is needed to map locations of restriction sites to the specific myosin heavy chain genes, as well as to specific regions inside the gene.

Statistical analysis
The differences between species and populations were tested by chi-square ( X 2 ) tests. In those cases where sample sizes were too small and the expected values of a cell were < 5, we used the Fisher exact probability test (Sokal and Rohlf, 1981). We doubled the calculated signifiance level for a 2-tailed significance test.

RESULTS
The results of hybridizing the MHC gene probe with S P alax DNA digested with several restriction enzymes are summarized in tables II, III and figure 2. In Spalax, as in other mammals, the MHC is a multigene family comprised of a minimum number of 6 loci, detected with the rat gene probe (see fig 2). We considered the number of large and strongly hybridizing bands obtained from digestion of 6-base-recognition endonuclease as the number of genes. The rationale for that consideration relies on: a) the fact that the 2 different probes yielded the same fragmentsie, there was no case where the endonuclease cleaved the DNA inside the probe homology region; and b) the MHC gene is much larger than the obtained fragments, so there is no reason to assume that one band corresponds to 2 linked genes. Six was the minimum number of strongly hybridizing bands.
A total number of 26 scoreable (on a population level) frequent fragments were produced by the 4 restriction endonucleases used in this study (table II).
There were some bands that could not be consistently detected in all samples, and were therefore excluded from the analysis. Most of the detected fragments are monomorphic in the 4 chromosomal species and in the 13 populations examined. Some of them, however, are polymorphic and are summarized in table III. Out of the 26 fragments, 8 (31%) presumptive loci were found to be polymorphic, with no more than 2 alleles observed per locus.
Codominance with alternative allelic fragments was found only in 2 loci. In the other 6 loci, we detected only the appearance or absence of the fragments.
We considered the missing alternative alleles as non-allele. The lack of alternative allele could result from several causes. First, the experimental conditions: small fragments could run out of the 0.6% agarose gels. Second, there is the possibility of dominance. This is unlikely in our case because the phenomeon is usually scarce yet absence of the fragment is common in our data set. Third, the possibility exists that gene duplication appears in some of the individuals tested. However, since the phenomenon appears across all 4 chromosomal species it must pre-date speciation. Short of family analysis, this possibility cannot be ruled out.
Lack of effective gene flow and differences in MHC polymorphism between the oldest species 2n = 52 and 2n = 54 An alternative fixation between 2n = 52 and 2n = 54 is evident at the KpnI markers. All samples of 2n = 52 were fixed for fragment III, while none occurred in 2n = 54. It is possible that fragment III does exist in 2n = 54 as a rare allele (with frequency lower than 0.1). The probability of finding no individuals with KpnI fragment III phenotype in a sample of 27 individuals is (1 -0.1)2! = 0.058. The differences between the 2 species were highly significant ( X2 = 55.98,p « 0.001).
These results suggest that there is no effective gene flow between the oldest chromosomal species 2n = 52 and 2n = 54. Similar results, though less pronounced, can be seen in the TaqI fragment XII. Most individuals of 2n = 52 showed this fragment which was rare or absent in 2n = 54. The difference between the 2 species in this locus was highly significant ( X2 = 39.31;p « 0.001). Some interspecies divergence is also evident in BstEII. In fragment II the average frequency of allele a is 32% in 2n = 52 and 85% in 2n = 54 ( X l 2) = 31.9,p << 0.001). The frequency of fragment V is 100% in 2n = 52 and only 48% in 2n = 54. Yet again the differences between the 2 species were highly significant ( X2 = 19.5,p « 0.001). These results suggest that there is either no gene flow between 2n = 52 and 2n = 54, or if there is, it is eliminated by selection.
Populations 1 and 7 of 2n = 52 and 2n = 58 straddle a hybrid !one, 10 km apart. Fragment V of BstEII was present in all individuals of population 1 (Maalot, 2n = 52) and in only 56% of the individuals from population 7 (Kabri, 2n = 58). The differences between the populations were not statistically significant (2-tailed Fisher exact test, p = 0.08) due to a small sample size. Similary, the variation in BamHI markers showed that the frequency of allele 6 (4.8 kb) in fragment IV was 25% in population 1 (2n = 52) and 0% in population 7 (2n = 58) (2-tailed Fisher exact test, p = 0.08). However, the combined probabilities from tests of significance between the 2 populations straddling the hybrid zone (Sokal and Rohlf, 1981) were significant ( X4 = -2E ln P = 10.04; p < 0.05). These results suggest that different selective pressure and possibly no effective gene flow operates across the hybrid zone separating 2n = 52 and 2n = 58.
These are 10 km apart, across a hybrid zone 0.7 km wide, which separates the species in the southern Golan Heights (fig 1) (Nevo and Bar-El, 1976;Nevo, 1985).
Three loci: BstEIIfragment II, BamHIfragment IV, and TaqI -fragment IX, were polymorphic in population 6 of 2n = 54 and monomorphic in population 9 of 2n = 58 (table III). These results may indicate that there is a possible lack of effective gene flow from 2n = 54 to 2n = 58, and possible different selective pressure that operate on these geographically close populations.
Limited amount of gene flow between the younger species 2n = 58 and 2n = 60 Reduction or absence of gene flow between the 2 younger 2n = 58 and 2n = 60 species is also suggested by several markers. Population 10 (Anza, 2n = 60) is located south of a 2n = 58 x 60 hybrid zone, while sample 8 (Zippori, 2n = 58) was collected from the center of the 2n = 58 range. BstEII fragment II variation showed 2 alleles in population 8 with a frequency of b at 56%, and the absence of this allele in population 10. The differences between the 2 populations were highly significant (2-tailed Fisher-exact test, p « 0.001). The opposite pattern was displayed by KpnI variation. The frequency of fragment III was 100% in population 8 but 56% in 10. Although population 8 is not located near the hybrid zone, these patterns suggest no widespread effective gene flow between the 2n = 58 and 2n = 60 populations.
Differences in MHC polymorphism within the 2n = 58 chromosomal species Population 9 (Afiq) of the 2n = 58 chromosomal species exhibits a pattern of variation strikingly different from the pattern of the other 2 populations. These differences appear in 3 out of the 8 polymorphic loci. The most distinguished pattern is in KpnI fragment III where the fragment fixed in populations 7 and 8 is totally absent in population 9 (2-tailed Fisher exact probability test, p «0.001).
Similarly, a quantitative rather than a qualitative pattern can be detected in the TaqI variation of fragment XII. The fragment is fixed in populations 7 and 8, and appeared at a frequency of only 33% in population 9. The differences between populations 7,8 and population 9 were statistically significant (2-tailed Fisher exact probability test, p <0.005). In BstEII fragment II variation, populations 7 and 8 are polymorphic with similar frequency of both allele a and b. Population 9 is monomorphic, consisting of only allele a (2-tailed Fisher exact probability test, p <0.02). The differences described may indicate the existence of a gene flow barrier between the populations.

Number of genes
Similar to other mammals examined (Nguyen et al, 1982;Leinwand et al, 1983;Maeda et al, 1987), we have recorded in the mole rat a minimum number of 6 large hybridizing fragments and a maximum of 14, corresponding to at least 6 but probably 8 genes. The complexity of the MHC pattern in southern blots is most probably an indication of the number of sarcomeric MHC genes in the organism (Nguyen et at, 1982). The complex structure of the rat MHC gene and the conservation of intron locations are indicative of a highly split ancestral sarcomeric MHC gene (Strehler et at, 1986). Once the split ancestral MHC gene evolved, it is likely to have undergone duplication(s) as an entire genetic unit in order to give rise to the multi-member sarcomeric MHC gene families (Strehler et at, 1986). The rat sarcomeric MHC gene family has diverged enough from smooth muscle and non-muscle MHC that there is no cross-hybridization between them (Nguyen et at, 1982;Leinward et at, 1983).
In the present analysis, we used rat DNA as a probe to the mole rat MHC genes. Although MHC appeared to be highly conserved during the evolution of straight muscles (Nguyen et at, 1982), we used medium stringency conditions of 2 x SSPE and 0.1% SDS at 63°to detect the Spalo,! genes. These conditions suggest that the homology between the rat and the mole rat is not very high. Further study is needed to determine both the exact number of genes and their characteristics.
The level of polymorphism Polymorphisms of some of the MHC genes were found with all the tested restriction enzymes. Although relatively high levels of polymorphism (31%) were found in MHC genes of Spalax ehrenbergi, the number of alleles identified was low. Table III shows that in the scoreable cases only 2 alleles were found. The differences between the observed and expected heterozygosity were not large, considering the small sample size. However, in BstEII fragment II variation, Ho were smaller than He in 8 out of 9 polymorphic populations (sign test, p < 0.04). These results may suggest an existence of selection operating against heterozygotes in this locus.
In other loci, we could not identify the alternative allele(s), though there is no reason to suggest the presence of more than one unidentifiable allele. Existence of a restriction site polymorphism was found in the rat (Nguyen et at, 1982), and in the human (Nguyen et at, 1982;Leinwand et at, 1983;Schwartz et at, 1986).
Although no population analyses have been conducted, it seems that in these species, as in Spalax, the number of alleles per locus is low. It is possible that genetic polymorphism in the MHC genes plays a role in generating the complex MHC patterns (Nguyen et at, 1982).
The low number of alleles in SPala! is similar to levels of isozyme polymorphism as determined by horizontal starch gel electrophoresis Shaw 1972, Nevo andCleve, 1978). The most polymorphic protein locus had only 3 alleles, and most of the other loci had only 2 alleles. In contrast, the major histocompatibility complex showed a very high number of alleles at the RFLP level (Ben-Shlomo et at, 1988).
In other words, while structural genes coding for basic metabolic enzymes and for heavy myosin chain are relatively conservative, those of the immune systems are extremely polymorphic. Low levels of allozyme polymorphism appear to be the general phenomenon in fossorial and subterranean mammals (Nevo et at, 1984, Nevo, Fillipucci, Capanna and Hickman, in preparation). In the myosin heavy chain, the level of polymorphism is ! 31% with few alleles (2) segregating at each polymorphic locus. Since equivalent population surveys on mammals living above ground have not yet been made, we cannot compare the level of polymorphism and the number of alleles between under and above ground mammals. Additional comparative studies are in order.
Differences in myosin heavy chain polymorphism between the chromosomal species RFLP variation in the sarcomeric myosin heavy chain multigene family allows one to differentiate among the chromosomal species, particularly among the oldest species 2n = 52,54 and 58 (table III). Similar results suggesting the possible lack of gene flow across the 2n = 52 x 54 and 2n = 52 x 58 borders were also found in analyses of major histocompatibility complex loci (Ben-Shlomo et al, 1988), rDNA nontranscribed spacer polymorphisms (Suzuki et al, 1987), and mtDNA haplotypes . The narrowness of the hybrid zones, the lower than expected proportion of hybrids, the nonrandom distribution of karyotypes (Nevo, 1985) and the genetic differences among the chromosomal species indicate that genetic isolation is established, and supports the case for species status of these races. Moreover, it suggests that natural selection operates against migrants.
Our results support an earlier conclusion (Nevo 1985) that reproductive isolation increases progressively northward. While reproductive isolation is still weak between 2n = 58 and 2n = 60, it becomes much stronger between 2n = 52 -54, and 2n 52 -58. Dispersal of hybrids into parental territories is restricted, presumably by a combination of genetic, cytogenetic, ethological, immunological and ecological incompatibilities (Nevo 1985), whereas hybrids appear inferior to the parental types.
Genetic differences within the 2n = 58 species Existence of a gene flow barrier between population 9 and populations 7 and 8 of 2n = 58 is a possible explanation of the differences and an alternative fixation among these populations. Population 9 (Afiq) is a marginal population that is located in the southern Golan heights (fig 1), south of the 2n = 54 x 58 hybrid zone. Geographically this population of 2n = 58 is surrounded by the Yarmouk river in the south-east (the political border between Israel and Jordan) and by the sea of Galilee and the Jordan river in the west. These geographical barriers may cause the isolation of population 9 in a small area and thus create an isolated relict. Alternative fixation of alleles between populations 7 and 8 and population 9 was found in Period gene homologous sequence (Ben-Shlomo et al, 1993). The unique pattern of population 9 can also be seen in mtDNA haplotypes  where some haplotypes exist only in this population. As shown in the previous section, there appears to be a possibility of no effective gene flow between the 2 chromosomal species 2n = 54 and 58. Nevertheless, KpnI variation showed that population 9 of 2n = 58 exhibits a similar pattern to that of the neighboring populations of 2n = 54, by being fixed for the none-III fragment, in contrast to samples 7 and 8 (Kabri and Zippori) of 2n = 58. This convergent genetic pattern may suggest a common selection pressure that operates similarly on all populations in the Golan heights. Such a selection pressure could be mediated for example by the type of soil (basalt) characterizing the area.
A possible operation of genetic drift on small isolated demes within the 2n = 58 cannot be dismissed, although it seems less probable. While genetic drift gradually increases the interpopulation variation of gene frequency, the genetic variability within population gradually declines (Nei, 1987). Since the proposed isolation of population 9 has probably existed for thousands of generations, the genetic diversity should be very low. In contrast to the expected low values, the diversity of this population is similar to the other populations of the same chromosomal species, in the MHC genes as well as in other genetically examined systems, such as the major histocompatibility complex (Ben-Shlomo et al, 1988) or rDNA nontranscribed spacer (Suzuki et al, 1987). Moreover, in Per-homologous sequence (Ben-Shlomo et al, 1993) and in the mtDNA haplotype  the number of different types in population 9 is higher than in the other 2 populations.
The considerations mentioned above strengthen the possibility that parallel genetic patterns due to balancing natural selection operate on the MHC genes in the Golan heights.

Polymorphism in isolated populations
Sede Boqer (population 13) is an isolated population in the northern Negev desert (fig 1). The 2n = 60 chromosomal species in the Israeli range persisted during 70 000 years in a continuous range connecting the coastal plains of the Negev and Sinai deserts with northern Africa (Nevo, 1989). This species became disjunct from its North African extension in the post-Wurm period (Lay and Nadler, 1972). The isolated population (13) studied here may have therefore been isolated > 10 000 years from the main range of 2n = 60. The separation from the main range by inhospitable kilometers of steppic and desert environment resulted in a low effective population size, of the order of only about 100 individuals (Nevo, 1989). This relictual situation and small effective population size seem to have lasted for thousands of generations. Nei et al (1975) have shown that the effect of population size reduction (bottleneck effect) on average heterozygosity is expected to last for hundreds of thousands of years after the recovery of population size. Maruyama and Fuerst (1985) claimed that while allele numbers showed significant excesses when the population rapidly expanded from a homoallelic state, population size contraction caused allelelic deficiencies. These neutral considerations would suggest that in the isolated population of Spalax, where the bottleneck effect was sustained for many generations, with no population recovery, the level of variability would be extremely low. The opposite pattern is displayed in the Sede Boqer isolate.
The results obtained in the present study indicate that there is no remarkable difference between population 13 (Sede Boqer) and the other population of 2n = 60. In the case of TaqI, fragment IX (table III), the isolated population is polymorphic while most of the examined populations were monomorphic. Notably, due to the small population size of this isolate, we analyzed only up to 4 individuals from this population, while in all other populations we analyzed = 10. More individuals from this isolated population may increase the level of polymorphism. Similar results were found in the Sede Boqer isolate in other genetic systems, for example the major histocompatibility complex (Ben-Shlomo et al, 1988) and several allozyme systems, mtDNA and chromosomal polymorphisms reviewed in Nevo (1989). Apparently, an extreme homozygosity is unlikely even in a very small and isolated population.
The variability found in several genetic systems is due presumably to a very strong balancing natural selection. If this isolate may ultimately evolve into a new species rather than become extinct, then it involves a relatively high level of genetic variation which may lead to successful speciation (Barton and Charlesworth, 1984;Carson and Templeton, 1984;1989).