OUP user menu

Hybridization between native and introduced species of deer in Eastern Europe

Aleksandra Biedrzycka, Wojciech Solarz, Henryk Okarma
DOI: http://dx.doi.org/10.1644/11-MAMM-A-022.1 1331-1341 First published online: 19 October 2012

Abstract

A consequence of introduction of alien species can be hybridization with a closely related native species. Hybridization can have a large effect on the genetic structure and conservation status of native populations. Here, we present a study of hybridization and introgression between native red deer (Cervus elaphus) and introduced sika deer (C. nippon) from 5 regions in Poland, the Kaliningrad District (Russia), and Lithuania. With a set of microsatellite loci and a mitochondrial marker, we uncovered extensive hybridization in all regions despite different population dynamics and no reports of hybrid individuals. We propose that sika populations in Eastern Europe were established with individuals coming from at least 2 different localities in southern Japan and eastern China. Legislation designed to reduce threats posed by sika deer could help to prevent further hybridization.

Key words
  • Cervus elaphus
  • Cervus nippon
  • hybridization
  • introduced species
  • introgression
  • invasive species
  • red deer
  • sika deer

The introduction of species outside of their natural range can have far-reaching and often harmful effects upon the biological diversity and functioning of invaded ecosystems and can cause significant losses in economic value of native natural resources. Introduced species are now considered to be the 2nd most important cause of global biodiversity loss after direct habitat destruction (Baskin 2002; McNeely et al. 2001; Wittenberg and Cock 2001). One important, direct consequence of species introduction can be hybridization with native species when they come into contact with species belonging to a closely related taxon (Arnold 2004).

Hybridization can have a large effect on the genetic structure and conservation status of native populations or species (Allendorf et al. 2001; Arnold 1997) and can cause their extinction (Rhymer and Simberloff 1996). Introgression of nonnative genes into the genome can increase the extinction risk of a native population by causing outbreeding depression (Edmands 1999; Gilk et al. 2004; Marr et al. 2002). Ongoing hybridization that results in admixture of parental genes from 2 different species and introgression (i.e., transfer of genes from one species to another by repeated backcrossing), can cause the native population to be replaced by one genetically and phenotypically resembling the nonnative form in some traits (Huxel 1999). However, individuals containing nonnative genetic material might be morphologically indistinguishable from those containing an intact native genome (Chan et al. 2006; Goodman et al. 1999; Weigel et al. 2002).

Here, we present a study of hybridization and introgression between native red deer (Cervus elaphus) and introduced sika deer (C. nippon) from regions in Eastern Europe where these 2 species co-occur. The native range of sika deer is in the Far East, including Japan, southeastern Siberia to eastern China (Manchuria), and Korea (McCullough et al. 2009). After 1860, when the 1st individuals were imported to Great Britain, the species was bred and introduced for hunting in many European countries (Lever 1977, 1985). In many cases, sika populations are small and local in distribution, but more extensive populations are established in Great Britain, Ireland, Germany, and the Czech Republic. Red and sika deer can interbreed in enclosures (Bartoš 2009; Harrington 1982) and in the wild (Diaz et al. 2006; Goodman et al. 1999; McDevitt et al. 2009; Senn and Pemberton 2009), although hybridization in the wild was only documented by molecular methods in Great Britain and Ireland (Goodman et al. 1999; McDevitt et al. 2009; Senn et al. 2010a; Senn and Pemberton 2009). The hybridization process at the molecular level has never been studied in Eastern Europe.

The 2 species exhibit many morphological and behavioral differences. Red deer are generally much larger than sika but exhibit substantial variation in body size. Western European individuals are slightly larger than sika deer (Senn and Pemberton 2009). Eastern European individuals reach body masses ranging from 125 to 210 kg for males and from 71 to 101 kg for females (Tomek 2002). The body mass for sika deer ranges from 103 to 151 kg for males and from 68 to 98 for females of the larger “hortuloid” subspecies, and from 48 to 86 kg for males and from 29 to 58 kg for females of the smaller “nipponoid” subspecies (Bartoš 2009). Estimates for the divergence dates between sika and Western European red deer based on a fossil-calibrated molecular clock range from 5.2 to 7 nrllion years ago (Ludt et al. 2004; Randi et al. 2001).

The 1st sika deer were brought to southern Poland in approximately 1895. Originally, the 7 imported animals were kept in captivity, but were subsequently released in the neighboring forests near Pszczyna. The 2nd documented introduction of 7 sika deer was in about 1910 in Kadyny forest, on the coast of the Baltic Sea in northern Poland (Bartoš 2009). According to historical records, the larger hortuloid type was introduced in Kadyny, whereas the smaller nipponoid type was introduced in Pszczyna (Matuszewski and Sumiński 1988). The southern population never reached high numbers and remains stable to date at about 20 individuals, whereas the northern one reached about 250 individuals in the mid-1960s, and expanded its range toward the east and south. Currently there are about 500 individuals in this area. Recently, sika deer are becoming increasingly popular farm animals in Poland (Solarz 2008).

In this study, we used molecular markers to evaluate whether hybridization between red and sika deer has occurred in Eastern Europe and to assess the current extent and distribution of gene flow between these 2 species. Because hybridization between red and sika deer has already been reported in natural populations elsewhere, we predicted that it also occurs in Eastern Europe, although it seems to be unnoticeable in the field and is therefore widely dismissed by hunters. In addition, we assessed the origin of the sika deer populations included in this study.

Materials and Methods

Sample collection.—Muscle samples of both species were collected between 2005 and 2009 from legally hunted deer. Hunters were asked to classify harvested individuals as red or sika deer and to note all cues that were not typical of each species, which might suggest a hybrid phenotype. A total of 176 red deer and 50 sika deer were collected from 5 loosely defined populations or regions. We sampled 49 red deer in Pszczyna forest (Polish southern population), 41 red deer in Kadyny forest (Polish northern population), 18 red deer in Kaliningrad District (Russia), and 39 red deer in Lithuania (Fig. 1). In all of these places, red deer and sika deer co-occur. Additionally, we sampled 29 red deer in an area located about 200 km southeast from the northern Polish sika location (Piska forest), because there were several observations of sika stags migrating in that direction. Sika deer samples come from one region in Poland (Kadyny forest, 46 individuals), and from Kaliningrad District (Russia, 4 individuals). We were not able to obtain sika deer samples from all regions where both species co-occur because harvesting of sika deer was not conducted at some places. We use the abbreviation E for red deer and N for sika deer throughout the text. All sampling sites are presented in Fig. 1.

Fig. 1

Map of the study area with sampling sites in Poland, the Kaliningrad District of Russia, and Lithuania. The distribution and frequency of mitochondrial DNA haplotypes is shown below the map. Red deer haplotypes are given in light gray and sika deer haplotypes in dark gray. The number of individuals bearing each haplotype is shown in parentheses. PL = Poland, LT = Lithuania, RUS = Kaliningrad District of Russian Federation, BY = Belarus, UA = Ukraine, SK = Slovakia, CZ = Czech Republic, D = Deutschland (Germany).

Microsatellite genotyping.—The DNA was extracted from ethanol-preserved tissue using a NucleoSpin Tissue Kit (Macherey and Nagel, Dueren, Germany) according to the manufacturer's protocol. We used 14 microsatellite loci to evaluate the extent of hybridization. Eleven of them (BM 757, VH64, OarFCB 193, INRA 6, RM 12, BOVIRBP, FSHB, RM 188, MM12, VH54, and TGLA40) were previously used in assessment of hybridization between red and sika deer by Goodman et al. (1999) and Senn and Pemberton (2009). The microsatellite loci used by these studies were selected for having no shared alleles between sika and red deer. This was tested in a panel of 44 sika and 44 red deer individuals taken from different locations in the United Kingdom. Additionally, we used 3 other loci (BL 42 and BM 203 [Bishop et al. 1994], and OarFCB 304 [Buchanan and Crawford 1993]) that proved to cross-amplify and were polymorphic in both species. The primers were originally developed for cattle (Bos taurus), except for OarFCB 193 and oarFCB 304, which were developed for sheep (Ovis aries). Genotyping of the loci used by Senn and Pemberton (2009) was performed according to their protocol. Three remaining loci were amplified in 10-µl reactions containing 10–20 ng of DNA, 0.2 µM of each primer, and 4.3 µl of Hot Star PCR Master Mix (Qiagen Ltd., Crawley, United Kingdom). Polymerase chain reaction amplification conditions were as follows: an initial denaturation at 95°C for 15 min; followed by 30 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 30 s; and a final extension at 72°C for 30 min. All polymerase chain reaction products were run on an ABI PRISM 3130xl Genetic Analyzer (Applied Biosystems, Foster City, California) and sized with the internal lane standard LIZ 500 using the program Genemapper version 4.0 (Applied Biosystems).

Mitochondrial DNA genotyping and sequencing.—Sika and red deer mitotypes were assigned for all samples using a 39-base pair (bp) tandem repeat in the mitochondrial control region (Cook 1993). Red deer have a single repeat, whereas sika deer have multiple repeats of this fragment. Length variation of the diagnostic fragment of the control region was assessed by electrophoresis in 4% agarose gel stained with ethidium bromide. Additionally, we obtained control-region sequences of a subset of individuals representing all populations, including all those that had discordant assignments to species based on phenotypic and control-region fragment size. Polymerase chain reaction products were purified and sequenced using both forward and reverse primers. Sequencing reactions were performed with Big Dye 3.1 sequencing kit (Applied Biosystems) and run on an ABI 310 Genetic Analyzer (Applied Biosystems). Sequences were checked and aligned with use of the ClustalX algorithm in BioEdit 7.0.5.3 (Hall 1999). All the haplotypes obtained have been deposited in Genebank (accession numbers HQ534296 and HQ534297 for sika deer and HQ534299–HQ534310 for red deer).

Genetic diversity analysis.—Estimates of genetic diversity of both species were obtained after exclusion of individuals identified as hybrids in the admixture analysis (described below). Basic locus-specific diversity measures (estimated separately for each species) such as allele frequencies and observed and expected heterozygosities (Ho and HE, respectively) were calculated using FSTAT version 2.9.3 (Goudet 1995). Because groups under comparison are not true populations, but rather species, they are not necessarily supposed to meet the Hardy–Weinberg expectation. Tests of linkage disequilibrium between all pairs of loci within each species and overall were implemented in GENEPOP version 3.4 (Raymond and Rousset 1995) using 10,000 permutations. Associated probability values were corrected for multiple comparisons using a Bonferroni adjustment for a significance level of 0.05. To compare the measures of genetic diversity between the species, we used a permutation test implemented in FSTAT. We calculated 1-sided probability values for the observed heterozygosity and gene diversity to test if they were higher in red deer. To compare the level of differentiation between species and populations within species, we computed pairwise FST between populations in FSTAT and preformed an analysis of molecular variance (AMOVA) using Arlequin 3.0 (Excoffier et al. 2005). The population of sika deer from Kaliningrad District was excluded from the analysis because only 1 individual was not identified as a potential hybrid.

Population admixture analysis.—Analysis of population and individual admixture using the microsatellite multilocus genotypes was carried out with a Bayesian clustering algorithm implemented in STRUCTURE 2.2 (Pritchard et al. 2000). This model assumes that there are K populations, each of which is characterized by a set of allele frequencies at each locus. Within populations, loci are assumed to be in Hardy–Weinberg equilibrium and independent. The model introduces population structure by assigning the ancestry of individuals probabilistically to 1 or more populations (Pritchard et al. 2000). This method generates estimates of the admixture proportion (Q) for each individual genotype in the sample set. A STRUCTURE analysis was chosen because assumptions about the population origins of alleles at each locus are not needed. Despite the fact that most of the loci used in this analysis shared no alleles between red and sika deer in the studies by Goodman et al. (1999) and Senn and Pemberton (2009), we cannot transfer this assumption to our study system because different species histories in continental Europe and on the British Isles could lead to different allele frequencies. In each case, STRUCTURE was run under the admixture model and correlated frequencies without using information on sample origin (Falush et al. 2003; Pritchard et al. 2000). The most likely number of populations in the data set (K) was estimated by conducting 10 independent replicates for each value of K between 1 and 8, using 100,000 final iterations after a burn-in period of 100,000 iterations. The most likely number of clusters (K) was calculated by obtaining the mean posterior probability of these data (In Pr(X/K)) over the 10 independent runs.

Each of the 224 individuals was assigned to either the sika deer cluster (when membership probability is 0 = Q = 0.05), red deer cluster (0.95 = Q ≤ 1), interspecific hybrid of red type (0.75 = Q ≤ 0.95), sika type (0.05 = Q ≤ 0.25), or hybrids of equal ancestry (0.25 = Q ≤ 0.75) that are probable 1st-generation hybrids. Choosing the threshold values for hybrid detection is always a critical point because it strongly affects the identification of hybrids and introgressed individuals. The most appropriate approach is performing simulation on parental genotypes to assess the power of markers used for hybrid identification.

To assess the power of the admixture analysis and to verify the selected threshold values, we performed hybrid simulation with Hybridlab 1.0 software (Nielsen et al. 2006). For our data, it was impossible to select a sample of pure parental genotypes of both species. We were not able to obtain a sample of pure sika deer genotypes because this species is in contact with red deer in all of our sampling locations. Obtaining a set of pure red deer parental genotypes also was problematic because we detected a hybrid individual in a location 200 km away from the closest wild sika deer population. To overcome this problem, we selected a set of 20 individuals of each species that were identified as nonadmixed (Q ≥ 0.95 for red deer and Q ≤ 0.05 for sika deer) in the preliminary run of STRUCTURE. These sets of individuals were used as parental populations to create a simulated number of 20 of each 1st-generation (F1) and 2nd-generation (F2) hybrids and lst-generation backcrosses (F1 × red deer and F1 × sika deer). Subsequently, a data set containing both real and simulated individuals was run in STRUCTURE to detect if all the simulated hybrids were detected at the defined threshold level. This approach allowed us to avoid identifying individuals that share alleles due to ancestral polymorphism as hybrids. The threshold of Q = 0.05 in STRUCTURE was able to detect hybrids in studies with 12 or more microsatellite loci when the FST-value was 0.21 (Vähä and Primmer 2006). Often, a cutoff of 0.01 is used (Beaumont et al. 2001; Chazara et al. 2010; McDevitt et al. 2009). We opted for a less-stringent threshold, which increased the chance of identifying hybrids. At the same time these more relaxed criteria may lead to identifying “pure” individuals as hybrids because of the presence of alleles that have persisted in the 2 species since their split ancestral polymorphism or due to convergent mutation of microsatellite allele lengths (i.e., alleles are identical by state but not by descent).

We implemented a model in program NewHybrids (Anderson and Thompson 2002) as a complementary approach to detect hybrids. This method uses Markov chain Monte Carlo sampling to determine deviations from Hardy–Weinberg equilibrium among multilocus genotypes to assess the posterior probability that individuals belong to each of the 6 genotypic classes that originate after 2 generations of hybridization (i.e., 2 parental types, lst-generation hybrids [F1, 2nd-generation hybrids [F2], and backcrosses of F1 with each of the parental species). In this analysis, as in STRUCTURE, no information from pure parental genotypes is required. Uniform priors in a 100,000 Markov chain Monte Carlo burn-in and 1,000,000 steps for sampling were used in 10 independent runs to assess the stability of the results. Individuals were assigned to 1 of the 6 genotypic classes if P ≥ 0.95, or to 2 or more genotypic classes if 0.95 ≥ P ≥ 0.05.

Mitochondrial DNA analysis.—We sequenced all samples for which the fragment length did not match the assignment to species based morphological data. Additionally, we obtained sequences for a number of individuals from all populations used in this study to assess population structure and to identify the origin of introduced sika individuals. Haplotype diversity was calculated in DnaSp version 5. (Rozas et al. 2003). Analyses of control-region data within each (morphologically defined) species excluded individuals possessing control-region haplotypes from the other species. Relationships among haplotypes from all sequences were examined using Network version 4.5 (Bandelt et al. 1999). To assess the structure of red deer populations based on haplotype diversity, we performed an AMOVA in Arlequin 3.0 (Excoffier et al. 2005). To infer the origin of studied sika deer populations, we performed a maximum-likelihood analysis including our sika deer haplotypes and haplotypes of Japanese origin from Nagata et al. (1999). Phylogenetic trees were reconstructed in TREEFINDER (Jobb et al. 2004). We used the FindModel application (Los Alamos National Security, LLC 2010) to select the model of nucleotide substitution that best fit the data using the Akaike information criterion. The selected model (HKY + G) was set as the model of sequence evolution for the TREEFINDER reconstructions, and the maximum-likelihood analysis was performed using the likelihood-ratchet method. Branch confidence values were estimated using the estimated likelihood weights approach (Strimmer and Rambaut 2002), in which values of ≥70% were considered good support for a clade (Hillis and Bull 1993).

Results

Genetic diversity.—Excluding individuals identified as hybrids in the admixture analysis, genotypes of 157 red and 34 sika deer were used to obtain estimates of diversity. We did not find any evidence for linkage disequilibrium between any pair of loci either in separate species or overall. Number of alleles per locus ranged from 6 to 24 in red deer and from 2 to 18 in sika deer (Table 1). The distribution of allele frequencies in sika deer was skewed, with only 1 or 2 alleles with relatively high frequencies, whereas allele frequencies in red deer were more evenly distributed. Locus-specific expected heterozygosity was higher in red deer (0.615–0.898) than in sika deer (0.083–0.694); observed heterozygosities exhibited a similar pattern (Table 1). One-sided permutation tests suggested that both values were higher for red deer at marginally significant levels (P = 0.05 for both comparisons).

View this table:
Table 1

—Locus-specific diversity measures estimated for the total sample (n = 191) of native red deer (E, n = 157) and introduced sika deer (N, n = 34), from 5 regions in Poland, the Kaliningrad District (Russia), and Lithuania. Individuals identified as hybrids were excluded from these analyses.

NaaNaaENaaNHEb totalHEbEHEbN
BL42242440.7110.8760.533
BM203141440.7150.8950.521
BM757171440.6610.8150.495
BOVIRBP9720.5160.6260.395
FCB1932119180.8030.8980.689
FCB3041313140.6540.8540.438
FSHB151450.7850.8610.693
INRA69640.3520.6150.083
MM0126640.5960.6170.561
RM12121250.4980.8210.162
RM188181280.7640.8110.694
TGLA408750.5600.6900.413
VH54161440.7390.8630.602
VH64161650.6510.8660.422
  • a Na = number of alleles.

  • b He = expected heterozygosity.

Genetic differentiation was highest for sika deer from the Kadyny forest relative to all other populations of red deer (FSt-values ranged from 0.334 to 0.366; Table 2). Pairwise FST-values between red deer populations revealed low, but mostly significant differentiation (0.026–0.064). An AMOVA revealed that 39.3% of the genetic variation was between species, 2.3% of variation was distributed among populations within species, and 58.4% of variation was found within populations. Permutation significance tests showed that variation at all levels was significant (P < 0.001).

View this table:
Table 2

—Pairwise FST-values for all population pairs of native red deer and introduced sika deer sampled from 5 regions in Poland, the Kaliningrad District (Russia), and Lithuania. Numbers of individuals used for calculations (n) are shown in the 1st row and significance levels are shown below the diagonal (an asterisk [*] indicates significance at the 0.003 level; NS indicates not significant at the 0.05 level). Population designations: 1 = red deer, Pszczyna forest; 2 = red deer, Kadyny forest; 3 = Piska forest; 4 = red deer, Kaliningrad District; 5 = red deer, Lithuania; 6 = sika deer, Kadyny forest.

123456
n443228153834
10.00000.03550.06040.04240.06430.3367
2*0.00000.02880.03060.05070.3396
3**0.00000.02650.04690.3607
4**NS0.00000.02940.3662
5***0.00000.3340
6****0.0000

Population admixture analysis.—The results of 10 independent runs of STRUCTURE simulations at each value of K produced consistent results, and a division of the data set into 2 clusters (K = 2) captured the greatest proportion of the structure with an average In Pr(X/K) of−23,356 (SD = 50.47). For values of K = 1, we obtained an average In Pr(X/K) of −27,356, and for K = 3–8, average In Pr(X/K) ranged from −23,876 to −26,257. For each individual, the proportion of ancestry from each cluster (Q) was obtained. Among 176 individuals phenotypically designated as red deer by hunters, 157 individuals had Q-values of 0.95–1.0 (“pure red deer”), 6 individuals had Q-values of 0.75–0.95 (hybrid of the “red type”), 7 individuals had values of 0.25 < Q ≤ 0.75 (intermediate hybrid), 2 individuals had Q-values of 0.05–0.25 (hybrid of the “sika type”), and 4 individuals had Q-values of 0–0.05 (“pure sika”). Among 50 individuals designated in the field as sika deer, 34 were genetically classified as “pure sika,” 3 fell into the “sika type” hybrid class, 2 were classified as intermediate hybrids, 2 were identified as “red type” hybrids, and 9 were classified as pure red deer. All individuals identified as putative hybrids are listed in Table 3.

View this table:
Table 3

—Results of admixture analysis for individual native red deer and introduced sika deer sampled from 5 regions in Poland, the Kaliningrad District (Russia), and Lithuania, identified as hybrids by at least 1 of 2 methods. Individuals are grouped according to sampling site and their identification in the field by hunters (E = red deer and N = sika deer). The mitochondrial DNA haplotypes are reported in GenBank under accession numbers HQ534296HQ534310.

Q-valueNewHybridsMitochondrial
(STRUCTURE)assignmentahaplotype
Kaliningrad0.766F2E
District E0.16F2 0.70 + N 0.30N
0.240F2E
Lithuania E0.409F2E
Pszczyna forest E0.006NE
0.015NE
0.382F2E
0.342NE
0.927EE
Piska forest E0.746F2 0.74 + E 0.26E
Kadyny forest E0.008NE
0.811F2E
0.852F2 0.73 + E 0.27E
0.852BX × E 0.3 + F2 0.3 + E 0.4E
0.003NE
0.938EE
0.592F2E
0.478F2E
0.336F2E
Kaliningrad0.997NN
District N0.997NN
0.98NE
0.926F2E
Kadyny forest N0.998EN
0.996EE
0.997EE
0.074NN
0.055NN
0.245F2 0.76 + N 0.24N
0.706F2N
0.881E 0.85 + F20.15N
0.994EN
0.956E 0.85 + F2 0.15N
0.966EN
0.720F2E
  • a Posterior probability of assigning individuals to 1 of 6 classes (2 parental types [E and N], 1st-generation hybrids [F1], 2nd-generation hybrids [F2], and backcrosses of F1 [BX ×] with each of the parental species); exact values are given only if lower than 0.95.

Hybridization was documented in all sites. In Kaliningrad District, of 18 individuals phenotypically assigned to red deer, we found 1 “red type” hybrid and 2 “sika type” hybrids. Interestingly, all 4 individuals that were identified in the field as sika deer were classified genetically as very close to red deer. Three of them were classified as “pure” red deer and 1 as a “red type” hybrid. Of 39 individuals of phenotypical red deer from Lithuania, only 1 was assigned to the intermediate hybrid class. In Poland, among 49 red deer sampled in the Pszczyna forest site, there was 1 “red type” hybrid, 2 intermediate hybrids, and 2 “sika type” hybrids. In the Piska forest, where no sika population is established, only 1 individual of 29 sampled was assigned to the intermediate hybrid class. In Kadyny forest, among 41 individuals assigned in the field to red deer, there were 4 “red type” hybrids, 3 intermediate hybrids, and 2 individuals assigned to the “pure sika” class. Among 46 individuals assigned phenotypically to sika deer, 3 appeared to be “sika type” hybrids, 2 were intermediate hybrids, 1 was a “red type” hybrid, and 6 were classified as “pure” red deer.

The analysis of admixture performed in NewHybrids mostly confirmed the results obtained by the previous method. NewHybrids classifies individuals into 6 different classes: pure red deer, pure sika deer, 1st-generation hybrids (F1), 2nd-generation hybrids (F2), and backcrosses of F1 with each of the parental species. All but 2 individuals classified in STRUCTURE as intermediate hybrids were classified as 2nd-generation hybrids in NewHybrids with a posterior probability higher than 0.90. Individuals previously identified as “red type” hybrids were classified as F2 individuals, or partly as F2 individuals and partly as pure red deer. Similarly, individuals classified as “sika type” hybrids fell partly into the F2 class and partly into the pure sika classification. More generally, there was strong concordance between the 2 assignment methods (Table 3).

Mitochondrial DNA introgression.—Mitochondrial DNA (mtDNA) introgression was rare and mainly from red to sika (5 of 6 cases), Q-values for those individuals were 0.720 (F2 in NewHybrids), 0.926 (F2 in NewHybrids), 0.997 (pure sika in NewHybrids), and 0.980 (pure sika in NewHybrids). Two of those individuals came from the Kaliningrad District population and 2 from the Kadyny forest. The single example of sika deer mtDNA introgression in an individual morphologically identified as red deer occurred at the Kaliningrad District population (Q-value of the individual 0.24, classified as F2 individual by NewHybrids). Four individuals classified in the field as red deer, but assigned to the pure sika class by STRUCTURE and NewHybrids, possessed the red deer mitochondrial haplotype, confirming their hybrid status. Of 9 individuals classified in the field as sika deer, but assigned to the red deer class in both Bayesian analyses, 6 possessed the sika deer mitochondrial haplotype, confirming their hybrid status, and 3 possessed the red deer mitochondrial haplotype, which raises questions about the correctness of species identification in the field.

Mitochondrial DNA diversity.—The number of mito-chondrial sequences obtained from each population along with the distribution of haplotypes between the sampling sites is presented in Fig. 1. Analysis of the control-region sequences revealed 12 haplotypes among 39 red deer sequences and 2 haplotypes among 26 sika deer sequences. The length of the sequence was 547 bp for red deer and 625 bp for sika deer. Overall nucleotide diversity (π), number of segregating sites (S), and haplotype diversity (HD) were 0.0088, 28, and 0.874, respectively, for red deer, and 0.0101, 39, and 0.159, respectively, for sika deer. The sika deer samples from the Kadyny forest and the Kaliningrad District were each monomorphic in their mtDNA. In red deer, 35.6% of the mtDNA variation was found among populations. However, there was considerable haplotype sharing among sites, as shown by a median-joining network (Bandelt et al. 1999; Fig. 2).

Fig. 2

Median-joining haplotype network of mitochondrial haplotypes of red and sika deer. Haplotypes 1 and 2 belong to sika deer. Numbers on interconnecting lines represent the number of mutational steps.,. Small black diamonds symbolize median vectors produced by the network software, representing missing or not sampled haplotypes.

To infer the origin of introduced sika deer in the study area, we constructed a maximum-likelihood phylogenetic tree (Fig. 3) including sika deer haplotypes from the present study and haplotypes of Japanese origin from Nagata et al. (1999).

Fig. 3

Maximum-likelihood phylogenetic tree of the sika deer haplotypes obtained in this study combined with data from Nagata et al. (1999) describing phylogeography of sika deer on the Japanese islands. Numbers on branches indicate support of ≥70% obtained by the estimated likelihood weights approach. The red deer haplotype (C elaphus Cell) was used as an outgroup.

Discussion

Hybridization between red and sika deer.—Our analysis of 14 hypervariable, unlinked microsatellite loci and an mtDNA marker in a sample of 225 red and sika deer from 5 different regions in Poland, Kaliningrad District, and Lithuania revealed ongoing hybridization between these 2 species across a large geographical scale. Possible genetic consequences of the presence of alien species include reduced fitness and disruption of local adaptation via the introduction of maladaptive gene complexes (Allendorf et al. 2001; Hutchings and Fraser 2008; McGinnity et al. 2003; Rhymer and Simberloff 1996). Examination of mtDNA data indicated that hybridization has occurred mostly between sika stags and red deer hinds. The same bias in hybridization was observed in Scotland (Senn and Pemberton 2009), and differences in body size between the species could explain these results. Attempts at mating between sika stags and red deer hinds also were observed in the field (Bartoš and žirovnicky 1982).

We found a total of 35 (15.5%) hybrid individuals in all regions studied, including the region where no sika deer population is established. This level of introgression is much higher than detected in a study performed on the British Isles (Goodman et al. 1999; McDevitt et al. 2009; Senn and Pemberton 2009). Goodman et al. (1999) described the hybridization process in Scottish populations as a bimodal hybrid zone, with deer falling into 2 distinct classes (red deer–like and sika-like). The authors concluded that occasional hybridization was followed by introgression through backcrossing into the 2 parental species, resulting in red deer and sika populations containing a large number of individuals with a small number of introgressed alleles. Senn and Pemberton (2009) estimated an overall 6.9% mixed ancestry except for 1 site where 43% of individuals were hybrids. A comparable study in Ireland (McDevitt et al. 2009) also revealed similar levels of hybridization, with 9.4% of hybrids among red deer–like individuals and 10.6% among sika-like ones.

Our results probably reflect higher levels of hybridization than those documented in Scotland and Ireland, but they also might be due to ancestral polymorphism of microsatellite loci used in this study. Eleven of the loci used in this study were used previously by Senn and Pemberton (2009) and Senn et al. (2010a). These authors showed that these loci were highly diagnostic with almost none of the alleles occurring in both species. In our study, we observed considerable differences in allele distribution between species. In red deer there were few alleles occurring in roughly equal frequencies. In sika deer, there were 1 or 2 predominant alleles, usually absent in red deer, and the remaining alleles were found in very low frequencies. Nevertheless, the level of allele sharing was higher than that detected in the Scottish population. The loci were chosen as diagnostic on the basis of sample individuals from the British Isles only, although the history of both species there differs significantly from that in continental Europe (Bützler 1986). Even though Great Britain is the country where sika populations are among the largest and most widespread in Europe (Bartoš 2009; Putman 2010), all present populations were established from individuals bred in captivity (Putman 2010). The red deer, although native to Britain, almost went extinct in the middle of 18th century due to deforestation and overhunting and most of the present populations come from multiple reintroductions from relict populations and deer parks (Perez-Espona et al. 2009). In contrast, the sika populations in our study seem to have been established by multiple introductions from Japan, as evidenced by mtDNA analyses. Contrary to Great Britain, the red deer populations in continental Europe are large and have not gone through a phase of strong decline (Bützler 1986). This would explain differences in allele frequencies relative to British populations and might have resulted in a higher level of shared ancestral polymorphism between species.

Among 35 individuals identified as hybrids, 10 were assigned to the intermediate hybrids class in STRUCTURE and to the F2 generation in NewHybrids. This contrasts with the results of Senn and Pemberton (2009), who found no intermediate hybrids in their study with the exception of 1 site. Taking into account that our sample size was much smaller than in the Scottish study, finding a high number of 1st-or 2nd-generation hybrids suggests that hybridization events are frequent and that crosses between hybrids occur. Furthermore, we found intermediate hybrids in all of our sample sites. Most of those individuals came from the Kadyny forest, where the density of red deer is low in comparison to other sites and the sika population is abundant and gradually expanding, despite hunting. There are observations of rutting sika stags emigrating outside the area of the highest population density. Moreover, sika stags are known to join red deer harems during the rut, and in such situations, they are ignored by red deer stags even when attempting to mate with red deer females (Bartoš and žirovnický 1982). This explanation also is supported by the finding of intermediate hybrids that were phenotypically identified as sika deer, but carrying a red deer mitochondrial haplotype. One intermediate hybrid was found in Piska forest, where no sika population is established. This site is located about 200 km east of Kadyny, suggesting that the risk of hybridization also is high in areas within the range of possible sika migration. Alternatively, sika deer that hybridized with red deer in this region could have escaped from 1 of several farms the area.

Interestingly, the number of hybrids with a red deer–like phenotype in Kadyny was almost double that in our sample from the the Pszczyna forest; this result is surprising because similar numbers of red deer were sampled from both populations and the sika population in the Pszczyna forest is small in contrast to the numerous red deer population. This finding does not support the suggestion that hybridization is promoted when the population of the introduced species remains small, that is, sika stags mate with red deer females that are more abundant. Unfortunately, we were not able to search for hybrids among individuals of the sika phenotype in this population, because sika deer were not hunted there.

The pattern of hybridization among individuals from Kaliningrad District was not completely clear because of limited sample size. Among 18 individuals identified pheno-typically as red deer, there were 3 hybrids, all of intermediate type, which would suggest a high intensity of hybridization. Among 4 individuals identified in the field as sika, 2 carried a sika mtDNA haplotype, but the analysis using microsatellite loci classified them as red deer. This pattern of introgression could be due to hybridization of sika females with red deer males at the early stage of contact between these 2 species, followed by a few generations of backcrossing with red deer.

From our data, it seems that it is difficult to identify hybrids on the basis of their external appearances. Body proportions (e.g., head, neck, and body size ratios) varied considerably among “pure” individuals within one population (Bartoš and Žirovnický 1982). Nevertheless, it is surprising that in our data set no hybrids were identified by hunters, especially in the light of the recent findings of Senn et al. (2010b). They found that hybridization between red and sika deer resulted in the increase in body mass of sika-like hybrids and the decrease in body mass of red deer–like hybrid females. The low levels of hybrid detection on the basis of the phenotype also might be due to the tendency of hunters to assign individuals to a pure form. Moreover, there are reports of producing intentional crosses of red and sika deer for trophy improvement of sika (Bartoš 2009), but these practices are usually secret. On the other hand, we also cannot exclude mistakes in sample assignment in the field.

Genetic diversity and structure based on microsatellite and mtDNA diversity.—Microsatellite diversity was significantly higher in red deer than in sika deer in our samples. This pattern is not surprising and likely reflects the contrasting history of these species in Eastern Europe. Red deer are widespread with high levels of connectivity among populations, whereas sika deer were introduced in relatively small numbers from limited sources.

>Similarly, we detected high numbers of red deer mtDNA haplotypes that did not show pronounced population structure, suggesting high migration abilities and no barriers to gene flow among eastern European populations. On the other hand, the number of haplotypes found in sika deer was low, as expected for a recently introduced species. We found only 2 haplotypes among 26 samples. All individuals from the Kadyny forest shared the same haplotype, and the 2nd haplotype was represented by individuals from the Kaliningrad District. Maximum-likelihood analysis revealed different origins of these 2 sika populations. The haplotype found in Kadyny groups with haplotypes from southern Japan, whereas the haplotype from the Kaliningrad District groups with haplotypes from Manchuria. The presence of the Manchurian haplotype in the population from Kaliningrad might be explained by the fact that the introduction of sika in the area of the former Soviet Union was conducted with individuals from the Russian Far East (Bartoš 2009). The finding that both populations were fixed for mitochondrial haplotypes suggests no exchange between them. Although the distance between these 2 populations is less than 200 km, they might be effectively isolated by a barbed wire fence that was erected along the Polish–Russian border after World War II. Nevertheless, a larger sample size would be necessary to critically evaluate this isolation.

Sika deer are considered to be among the most invasive introduced species in Europe (DAISIE 2009), and it has been argued that immediate actions should be taken to prevent the potentially rapid expansion of this species (Nentwig et al. 2010). Although red deer and sika deer in all regions we studied have been in contact for a relatively short period of time (around 150 years), the hybridization process seems to be extensive, possibly causing threats to native red deer populations. Control of hybridization through fencing in these regions is practically impossible, because the sika populations, except for the one in Pszczyna, are numerous and distributed over large areas. For the same reasons complete culling also seems to be impossible; moreover, attempts to significantly reduce population numbers are not desirable from a conservation point of view, because heavy, but incomplete culling can trigger or intensify the hybridization process. In any case, attempts to confine or reduce wild sika populations would be hampered because of frequent escapes and releases from farms. However, more effective legal frameworks reducing threats posed by sika deer are necessary. No new intentional introductions into the wild should be authorized. Keeping and breeding of animals should only be allowed when risks of escape are minimized.

Acknowledgments

This work was supported by a grant from the Polish Ministry of Sciences (N303/P01/2007/32) and EU FP 6 Integrated Project ALARM: Assessing LArge-scale environmental Risks for biodiversity with tested Methods (GOCECT-2003-506675). We thank all the people who contributed to sample collection, especially L. Balciukas, who provided samples from Lithuania, W. Bragiel from Kobiór Forest Inspectorate, W. Tylkowski from Kadyny Forest Inspectorate, J. Bobek from Zaporowo Forest Inspectorate, and E. Kozlowski, who provided samples from Kaliningrad District. We thank M. Konopiński for his help in phylogenetic analysis and comments on the manuscript. Special thanks to E. Lessa, who helped with language corrections and gave essential comments on the manuscript.

Footnotes

  • Associate Editor was Janet. A. Rachlow

Literature Cited

View Abstract