Genetic depletion within populations and increased genetic differentiation among them should be quite obvious when populations are kept in enclosures over a number of generations. Such captive populations were usually founded with small numbers of individuals and allowed to grow up to population sizes frequently corresponding to an effective population size (N_e) ranging from 50 to 100, at an annual culling rate roughly equalling the annual rate of population increase [6]. In long-term studies on captive wild boars allele frequencies were shown to fluctuate considerably under such conditions [7]. Dramatic changes in allele frequencies may also be spotted by comparing genetic relatedness among populations with their respective geographic locations. Allele frequencies may depart from natural conditions due to the small number of founders, the mixed geographic origin of founders, and the subsequent fluctuations from generation to generation. When a sample of 17 red deer populations, seven of which were living in enclosures, was subjected to phylogenetic analyses, trees based on genetic distances among populations were topologically very different from those based on geographic distances among sampling sites. However, when the enclosures were excluded, a much better match of genetic and geographic tree topologies was achieved [8].

In terms of reduction of genetic variability it is usually not average heterozygosity, but the proportion of polymorphic loci and the mean number of alleles per locus which are most affected by a small N_e in the absence of migration [9]. Altogether, in captive red deer populations studied so far significant losses of genetic variability (at least in terms of allozyme variation) do not seem to be a major problem [8]. Apart from founders coming from different source populations [10] this may be due to the fact that rare allozyme alleles in red deer are scarce anyway. Most polymorphic loci contributing to genetic variability show two major alleles occurring at high frequencies which make the loss of one or the other allele a rather unlikely event.

The biogeographic history of European red deer has been under human influence since very long ago, possibly since ancient times [2]. During the last one thousand or so years there has been an extensive trade in this species aiming mainly at the improvement of trophy quality [11]. As a consequence, extinct or nearly extinct populations have been restocked and autochthonous populations have been hybridized with introduced animals, thus blurring the historical and genetic boundaries between formerly natural populations. For example, in Italy the only remaining autochthonous stock is the small population of Mesola Wood in the Po delta [12]. Similarly, most British populations are allochthonous as well [2]. Apart from single hybridizations with American wapitis and central Asian subspecies there have mainly been translocations within Europe. What is known about the extent of these translocation and restocking activities is – due to a lack of documentation – only the tip of the iceberg, but even so, it is obvious that human impact must have been immense [1,2]. One example of human intervention is the allochthonous north-west Italian population in Val di Susa: It was founded in the 1960s with ten specimens, seven of them being of Slovenian and three of them being of Bulgarian origin (Apollonio, pers. comm.). The genetical similarity between Val di Susa and Bulgarian red deer is still obvious: They share an exclusive mitochondrial haplotype as well as some rare microsatellite alleles. Besides, among the populations under study Val di Susa showed the lowest genetic distance to Bulgaria based on microsatellite allele frequencies [13]. Red deer have been introduced to islands by humans as well as was the case on the Isle of Rhum off Scotland [14] and as has also been suggested for Cervus elaphus corsicanus which inhabits Corsica and Sardinia [15–17]. As to this particular subspecies, however, the hypothesis of artificial settlement is questionable since red deer might well have reached Corsica (and from there also Sardinia) via mainland Italy during the last glaciation. This hypothesis probably fits paleontological findings best. It is based on microsatellites but is in clear contrast to data based on mtDNA sequences [13]. Microsatellites have also been used to uncover genetic differentiation in Japanese sika deer (Cervus nippon) and to identify the origin of sika deer introduced to the United Kingdom [18]. Enclosures seem to have played an important role in transferring and restocking of red deer as they served as reservoirs of different populations and subspecies of red deer, let alone other deer species (cf. Woburn in Bedfordshire which is the origin of feral populations of four different exotic species of deer) [2]. Interestingly, the translocations of wapitis and Asian red deer as well as of the famous east European stags from Rominten were by far not as successful as had been hoped for with regard to the improvement of antler trophies [1]. This failure is very probably due at least partly to the lack of adaptation to local environmental factors as can be seen from the high susceptibility to disease in wapitis transferred to Europe [1]. In some cases a species' genetic integrity can be threatened by the translocation of a different species (although, of course, this conflicts with the biological species concept). This holds true for British red deer which readily hybridize with introduced sika deer to such an extent that red deer populations in north-west England [19] and Scotland [20] are in danger of losing their specific identity.

Based on hunting legislations dating back to 1935, in several European countries yearlings with spike length not exceeding ear length are selectively eliminated from populations and so are stags two to five years old with a low number of antler points. The underlying assumption is that both phenotypes are firmly associated with a later development of small and poorly branched antlers [21]. Two questions emerge: (1) Does this hunting practice increase the frequency of large and branched antlers in a red deer population? (2) If it actually works, should this practice be applied to a red deer population or must it be expected to be dangerous for long-term survival? Particular genotypes at the enzyme loci Idh-2 and Acp-2 were found to be significantly correlated with more branched and generally larger antlers, respectively [6,21]. Results suggested these allozyme loci to be linked with two independent genetic components affecting antler growth, one (Idh-2) being particularly relevant in young males, the other (Acp-2) rather in stags with fully developed antlers (i.e. males older than seven years). It could be shown in an isolated population (Vosges du Nord, France) that selective hunting significantly altered the frequencies of alleles at the two marker loci in the expected way [21]. At Idh-2 the same alleles associated with antler growth were in certain genotypic combinations and, in part, with age-specific differences shown to be positively correlated also with juvenile survival and female fertility [22,23]. Thus, albeit somewhat indirectly, it can be concluded that selective hunting may well have an effect not only on the traits in question but also on several characters associated with fitness. Selective hunting for antler shape and size should thus definitely be abandoned.

In Central Europe, concern as to the possible genetic consequences of habitat fragmentation and hunting legislations calling for an isolation of populations was expressed as early as 1976 [24]. Blood group analyses [25], electrophoretic transferrin analyses [26], and multilocus allozyme screenings [27] confirmed expectations of small and isolated populations harbouring less genetic variation than big ones. In a first screening of European red deer using mtDNA-RFLPs [17] several populations in France appeared to be monomorphic for one or the other out of a pool of several haplotypes. To analyse genetic differentiation among French red deer populations in more detail a total of 472 red deer sampled from 16 populations (Fig. 1) in 1992 and 1993 were studied for electrophoretic variation at seven enzyme loci found polymorphic in previous studies [28]: Me-1, Me-2, Idh-2, Sod-2, Acp-1, Acp-2, and Mpi. Allele frequencies were significantly different among populations (p<0.001); overall F_ST was 15%. Within populations there was no deviation of observed genotype frequencies from Hardy–Weinberg equilibrium (HWE). When all populations were pooled, a statistically significant excess of homozygotes occurred at all loci (Wahlund effect). F_ST among the populations in the Paris region only (Chantilly, Compiègne, Ermenonville, Halatte, Retz) was 7% and allele frequencies at all loci were significantly different. F_ST for the southern study area (Brouard, Champchévrier, Marchenoir) was 9%, and allele frequencies of six out of seven loci were significantly different. F_ST in the Vosges (Donon, Gérardmer, Vosges du Nord) was 3% and only two out of six loci showed significant allele frequency differences. When the populations of the respective subsets were pooled, there was no significant deviation from HWE at most loci. Interestingly, in the Vosges the only loci not being in HWE were Idh-2 and Acp-2. In both cases there was an excess of both possible homozygotes, which corresponds surprisingly well with the result that allele frequencies at these genes are influenced by selective hunting [21]. Genetic distances were not correlated with geographic distances when all populations were considered (Mantel-test, t=0.53, p=0.702, r=0.075). To avoid a bias caused by spurious genetic distances [29], all but one population of each of the aforementioned subsets were excluded in a second analysis, which did not change the result. In 73 specimens from 14 populations mtDNA-RFLP-analyses were carried out as described previously [17]. Five haplotypes were detected. Their distribution is shown in Fig. 1.

Altogether all parameters in our results demonstrate considerable heterogeneity among red deer populations across France. Except for deviations from HWE this holds true also for the Paris and the southern subset. This might well be due to habitat fragmentation caused by motorways or other barriers, but since we do not have detailed information on population histories, local effects of translocations cannot be ruled out either. The Vosges populations are quite homogeneous throughout our study area, which is in accordance with their autochthonous state and their demographic history [6,21].

Most of the body of work available to date on anthropogenic influences on the gene pool of red deer populations has been accumulated by allozyme electrophoresis [8,17,28,30]. In this marker system there are a few very ubiquitous polymorphisms for two major alleles (loci Me-1, Me-2, Idh-2, Acp-1, Acp-2). Particular Idh-2 genotypes appear to be associated with a number of morphological and fitness parameters [6,21–23]. Acp-2 and Me-1 genotypes are associated with antler development [6,21]. There are but a few loci which are only occasionally polymorphic (Sod-2, Mpi, Gpi-1). In our opinion it seems justified to ascribe the maintenance of most of the present polymorphisms in red deer to natural selection. Thus, average heterozygosity assessed using allozymes is probably not very representative of overall genetic variation. Typically, contrary to, for example, the white-tailed deer with many loci polymorphic for many alleles [31], average allozyme heterozygosity in red deer was found to be neither associated with body and antler growth nor with fluctuating asymmetry and fitness parameters [6,21–23].

Because of many more alleles detected at single loci and alleged selective neutrality microsatellites appear to be very suitable indicators of overall genomic variability. In red deer, microsatellites and microsatellite specific parameters taking into account quantitative differences among alleles showed individual molecular variation to be associated with birth weight and neonatal survival [32,33].

As regards geographic differentiation and phylogenetic relationships among deer populations microsatellites also turned out to be a powerful tool [13,18]. When the information was compared with data gained from mtDNA analyses, the results were sometimes corroborated [18], sometimes contradicted [13] by the maternally inherited system. Gene trees not matching separation of populations or species [34] and sex-specific migration or translocation patterns may account for these findings. At a European scale mtDNA-RFLPs based on some 70 restriction sites, corresponding to indirect sequencing of about 2% of the total mtDNA, showed a resolution power high enough to uncover patterns of genetic differentiation (Fig. 1, [17]). However, when compared with sequence data of the mtDNA control region, which is highly informative both at the infra- and the supraspecific level [35,36], results are not always concordant [13].