Genetic variability and differentiation in red deer (Cervus elaphus L) of Central Europe

A total of 365 specimens of red deer (Cervus elaphus L) from France, Hungary and Austria were examined for genetic variability and differentiation at 34-43 isoenzyme loci by means of horizontal starch gel electrophoresis and enzyme specific staining procedures. Calculated over 34 loci, mean P is 11.4% (SD, 2.05%) and mean H is 3.5% (SD, 0.8%). These values are similar to those detected in other studies on red deer. Relative genetic differentiation (Gg T) is 10.4%, but absolute genetic distances are very small throughout the study area, suggesting that all populations belong to the same subspecies (C e hippelaphus). In populations living in enclosures no general decrease in genetic variability was detected, but some rare alleles may have been lost and allele frequencies seem to be altered by genetic drift. The most ubiquitous polymorphism is that in IDH-2, which, together with other evidence, suggests that it may be maintained by selection. Other common polymorphisms are those in ME-1, ACP-1 and ACP-2, whereas variation in MPI, GPI-1, LDH-2, PGM-2 and SOD-2 shows a scattered distribution. Aspects of local differentiation among populations in France, Hungary and Austria are discussed. red deer / electrophoresis / isoenzymes / genetic variability / genetic distance Résumé-Variabilité et divergence génétique chez le cerf rouge (Cervus elaphus L) d'Europe. La variabilité électrophorétique de 34-43 locus enzymatiques a été examinée chez 365 cerfs nobles (Cervus elaphus L) représentant 17 populations originaires de France, de Hongrie et d'Autriche. Les taux moyens de polymorphisme (P) et d'hétérozygotie (H) sont respectivement de 11,.!% (t 2,05%) et 3,5% (t 0,8%). La différenciation génétique relative moyenne (G ST) est de 10,.j%, mais les distances génétiques absolues sont très faibles sur la région examinée, suggérant ainsi que toutes les populations analysées appartiennent à la même sous-espèce (C elaphus hippelaphus). Aucune réduction notable de diversité généti-que n'est observée pour les populations vivant en enclos, mais quelques allèles peu fréquents * Correspondence and reprints ont été perdus par certaines d'entre elles et la fréquence de certains autres allèles semblent avoir été modifiée par la dérive génétique. Le polymorphisme, le plus commun est celui d'IDH-2. Le caractère ubiquitaire du polymorphisme observé à l'isoenxyme IDH-2 conforte l'hypothèse, émise à partir d'autres arguments, d'un maintien de ce polymorphisme par sélection. On constate que certains polymorphismes, comme celui de ME-1, ACP-1, ACP-2 sont très fréquents, tandis que d'autres comme ceux de MPI, OPI-1, LDH-2, PGM-2 et SOD-2 montrent une distribution plus …


INTRODUCTION
Deer are among the few groups of large mammals which have been extensively studied by electrophoretic multilocus investigations during the last decade to evaluate genetic diversity within and between populations and species (see Hartl and Reimoser, 1988 for review). In red deer (Cervus elaphus L), biochemical genetic studies were carried out mainly in Scottish (C e scoticus L6nneberg, 1906;Dratch, 1983, Gyllensten et al, 1983Dratch and Gyllensten, 1985;Pemberton et al, 1988), but also in Swedish (C e elaphus L) and Norwegian (C e atlanticus 'L6nneberg, 1906) populations (Gyllensten et al, 1983). Genetic divergence between Scottish and American (C e canadensis Erxleben, 1777) red deer was examined by Dratch and Gyllensten (1985). Values of polymorphism and average heterozygosity estimated in all these studies are within the range generally observed in mammals (Baccus et al, 1983;Nevo et al, 1984). Population genetic studies in European red deer have also been carried out in Germany and Hungary in demes of the local form C e hippelaphus Erxleben, 1777 (Bergmann, 1976;Kleymann, 1976;Albert, 1984;Bergmann and Moser, 1985;Herzog, 1986;Kabai, 1987;Herzog, 1988a,b). However, due to the small number of loci or to the restricted geographical origin of the individuals examined, no overall values of genetic diversity could be calculated for comparison with the data given on other subspecies in the papers mentioned above.
To obtain a comprehensive picture of biochemical genetic variation and differentiation in red deer of Central Europe and a basis for comparison with other morphological subspecies, we conducted an electrophoretic investigation of 34 to 43 loci in various red deer populations from France, Hungary and Austria, which belong to C e hippelaphus (Wagenknecht, 1986).
The interpretation of electrophoretic band-patterns was carried out following the principles of Harris and Hopkinson (1976) and Harris (1980). Since no family studies could be performed, to reduce the possibility of misinterpretation samples containing enzyme variants were prepared once again and submitted to repeated electrophoretic runs. Furthermore, the results were compared to genetic variation in deer as described by other authors (see table IV for references) and to that detected in deer and other mammals in our laboratory, where the same enzyme systems were investigated (eg Hartl, 1986bHartl, , 1987Hartl, 1986, 1987;Hartl and Csaikl, 1987;Hartl and Reimoser, 1988;Hartl et al, 1988a;Leitner and Hartl, 1988;Hartl et al, 1990a). The number of genetic loci determining the various isoenzyme systems was assessed by comparison with data on deer species found in the literature (Table IV) and results from studies on the biochemical systematics of Artiodactyla and the homology of isoenzyme loci among mammals (see Hartl et al, 1988bHartl et al, , 1990b. Isoenzymes (and the corresponding gene loci) were assigned with numbers from the most cathodally to the most anodally migrating fraction. The most common alloenzyme (and the corresponding allele) in population BK * (Fig  1) Ayala (1977). We also calculated the average gene diversity within subpopulations (H s ), the total gene diversity (H T ), the average gene diversity among subpopulations (D ST ) and the relative magnitude of gene diversity among subpopulations (G ST ) according to Nei (1975). To examine the absolute genetic divergence among populations, several distance measures as compiled by Rogers (1986) were applied. Since the results obtained by using these different distance measures were very similar (as to be expected for small distances at the population level), only Nei's (1972) standard genetic distance and its version including a correction for small sample sizes (Nei, 1978) are presented in this paper.
To examine biochemical genetic relationships among the red deer samples studied, dendrograms were constructed by various methods (rooted and unrooted Fitch-Margoliash tree, Cavalli-Sforza-Edwards tree, Wagner network, UPGMA; see Hartl et al, 1990b) using the PHYLIP-programme package of Felsenstein (see Felsenstein, 1985). Since in some of the dendrograms only distances can be used which fulfill the triangle inequality, Rogers distances were chosen in these cases.
To test the influence of sample size and the composition of genetic loci chosen, the bootstrap and the jackknife methods were applied (for the use of the bootstrap in phylogeny, see Felsenstein, 1985). For the bootstrap, all observed allele frequencies are used to simulate new frequencies according to the sample sizes of the various demes studied. For the jackknife, 25% of the gene loci are randomly omitted. In each method, 100 new data sets are generated and used to construct phenograms, which form the basis for a consensus tree. The latter displays the most stable clusters and in a comparison with the original tree the weak points in the data become visible.
Allele frequencies are given in table I, values of polymorphism, heterozygosity and average heterozygosity are shown in table II. Genetic distances, corrected for small sample sizes (Nei, 1978) When all sampling sites are included, the dendrograms based on various distance measures and constructed by different cluster algorithms show only poor agreement with the geographic distributions of samples (see eg fig 2). It must be considered, however, that nearly half of them are quite small populations living isolated in enclosures, where extensive allele frequency changes, due to the founder effect, inbreeding, genetic drift, hybridisation of red deer from different provenances (see eg Bergmann, 1976;Kleymann, 1976;Leitner and Hartl, 1988;Hartl, 1989) and possibly also selection (Hartl et al, in preparation), are to be expected. Therefore the dendrograms were recalculated for only the free-ranging demes and in this case, except for changes of little significance, the topology among all of them was identical and showed a fairly good agreement with the geographic distribution of sampling sites (examples are shown in Figs 3 and 4). The stability of the main clusters is also demonstrated in a jackknife (Fig 5) and a bootstrap (Fig 6) consensus tree.

DISCUSSION
The genetic variability detected in the present study with mean P = 11.431 (SD 2.05%) and mean H = 3.5% (SD 0.8%) is similar to that obtained in previous investigations in populations of Scotland and Northern Europe (Gyllensten et al, 1983; see table IV). Since the sample of biochemical markers studied by these authors is quite similar to ours, regarding both the number and the composition of enzyme systems, the Pand H-values obtained in both investigations are thoroughly comparable. Our data are less comparable to those of Herzog (1988b) for 2 reasons. One is the very different number of genetic loci studied. The second is that his data are not based on a random sample of proteins, because the set of enzymes examined by Herzog (1988a), where no polymorphism was detected in pure red deer, was simply extended by including proteins, which were already known to be polymorphic in German red deer (Bergmann, 1976;Gyllensten et al, 1983). Therefore, as the author states himself, his data may give an overestimation of overall genetic variability (mean P = 13%, mean H = 3.4%). The problem of the influence of the number and composition of proteins studied can be demonstrated also by our data. If in the French red deer the polymorphism in ACP-1 is included in the calculation of genetic variation, the Pand H-values rise to mean P = 13.3% instead of 11% and mean H = 3.9% instead of 2.9%, but this is fully compensated by the 8 monomorphic loci investigated additionally in these populations (mean P = 10.6%, mean H = 3.1%). As stated by Gorman and Renzi (1979), slight differences in Pand H-values should therefore not be overemphasized. Bearing this argument in mind, a comparison of genetic variability within populations of the more extensively studied deer species (table IV) shows that mean Pand Hvalues as well as Pt calculated for the species (by summing up all polymorphic loci in the various demes) are rather high in the white-tailed deer, the reindeer and the roe deer, followed by the red deer. In contrast, the fallow deer and the moose exhibit considerably lower mean values of polymorphism and heterozygosity; in the fallow deer likely (Pemberton and Smith, 1985;Randi and Apollonio, 1988), in the moose possibly due to past genetic bottlenecks (Ryman et al, 1977. The polymorphism at the Idh-2 locus has been found (with the exception of Herzog, 1988a,b) in almost all red deer populations studied so faracross several subspecies (C e elaphus, scoticus, germanicus, hippelaphus and canadensis) -with high frequencies of (most likely) the same variant allele, ldh-2 12 , 1 (Dratch, 1983;Gyllensten et al, 1983;Dratch and Gyllensten, 1985;Hartl, 1986). The ubiquitous distribution of this polymorphism can be possibly explained by natural selection, when the results of Pemberton et al (1988), obtained in a red deer population on the Scottish island of Rhum, are considered. These authors detected significant associations between genotypes at the loci Idh-2, Mpi, and Tf (the variation at each of them is represented by 2 alleles) and juvenile survival. In Idh-2, heterozygote female calves survive much better than homozygotes, whereas male homozygotes survive better than heterozygotes, and the difference in survival is smaller. It cannot be decided if selection acts on the enzyme locus itself or at