Mitochondrial DNA variation and the phylogeography of the grey partridge (Perdix perdix) in Europe: from Pleistocene history to present day populations

Authors


 Tuija Liukkonen-Anttila, University of Oulu, Department of Biology, PO Box 3000, Fin-90014 Oulu, Finland. Tel.: +358-8-5531220; fax: +358-8-5531061; e-mail: tuijala@mail.student.oulu.fi

Abstract

Abstract For a phylogeographical analysis of European grey partridge (Perdix perdix) we sequenced 390 nucleotides of the 5′ end of the mitochondrial control region (CR) of 227 birds from several localities. The birds were divided into two major clades (western and eastern) which differed in control region 1 (CR1) by 14 nucleotide substitutions (3.6%). For estimation of the time of divergence, the whole CR of 14 specimens was sequenced. The major clades differed by 2.2%, corresponding to an estimated coalescence time of c. 1.1 million years. On CR1, 45 haplotypes were found. Western clade haplotypes were found in France, England, Germany, Poland, Italy and Austria. Eastern clade haplotypes were found in Finland, Bulgaria, Greece, and Ireland. One Finnish population and all Bulgarian and Irish populations were mixed, but only in Bulgaria was the mixing assumed to be natural. Nucleotide and haplotype diversities varied between populations, and both clades showed geographical structuring. The distribution of pairwise nucleotide differences in the eastern clade fitted the expectations of an expanding population. About 80% of the genetic structure in the grey partridge could be explained by the clades. The western clade presumably originates on the Iberian Peninsula (with related subtypes in Italy), and the eastern clade either on the Balkan or Caucasian refugia. Large-scale hand-rearing and releasing of western partridges have introduced very few mtDNA marks into the native eastern populations in Finland.

Introduction

Several animal and plant species recolonized Northern Europe from at least two different glacial refugia (Taberlet et al., 1998; Hewitt, 1999; Hewitt, 2000). One part of the fauna and flora arrived from south-western parts of Europe; another part dispersed from the less-well-studied eastern or south-eastern refugia (Hewitt, 1999, 2000). The populations in these refugia may represent a much older divergence than that created by the Pleistocene glaciation (Avise & Walker, 1998).

The grey partridge (Perdix perdix) is a species of grass steppe and cultivated farmlands of temperate climate. Its distribution range covers large areas in Europe and Asia, from Ireland to the Ural Mountains, excluding Iceland and the northernmost parts of Fennoscandia and Russia (Fig. 1). The world-wide decline in the number of grey partridge is well documented (for review, see Potts, 1986). A marked decline in the distribution range has occurred during the last century, mostly beginning in the 1950s as a result of modern agricultural practices. The species is on the brink of extinction at least in Ireland and Switzerland (Aebischer & Kavanagh, 1997).

Figure 1.

The European distribution area (shaded area) of the grey partridge ( Perdix perdix ) based on Potts (1986) , and Aebischer & Kavanagh (1997) . For abbreviations, see Table 1 . Pies describe the proportion of western and eastern haplotypes found in the analysed populations. Small black dots indicate additional sampling locations. Arrows indicate possible post-glacial recolonization routes for the western and eastern clade.

In Finland, the grey partridge lives at the edge of its northernmost distribution range (Potts, 1986) and is classified as a near-threatened species (Rassi, 2000). The population has decreased from 15 000 pairs in the 1950s (Merikallio, 1958) to approximately 4000 pairs in the early 1990s (Koskimies, 1992). According to Kivirikko (1948) the grey partridge invaded Finland in the beginning of the nineteenth century from the south-east, but according to old church parish records from southern Finland, it may have entered as early as in 1690 (Merikallio, 1958). The species was first introduced for hunting purposes in the middle of the eighteenth century (Merikallio, 1958) with birds most probably imported from Sweden (Kreuger, 1950). The Finnish population is suggested to represent subspecies P. p. lucida, whereas the nominate P. p. perdix is a more southern and western subspecies (Potts, 1986). Subspecies have evolved as a consequence of isolation and genetic adaptation of populations to local conditions.

Many releasing programmes have been carried out to strengthen natural populations of the grey partridge. However, several studies have clearly shown that the survival of hand-reared birds is poor after their release into the wild (Kavanagh, 1998; Putaala & Hissa, 1998). Siivonen (1957) suggested that differences in the genetic adaptation of separate subspecies (P. p. lucida, P. p. perdix) to different climatic conditions of their original range could be a partial explanation to the failed introductions in Finland. Maladaptive traits may be introduced into populations when hand-reared animals are released into the wild.

We analysed variation in the mtDNA control region (CR) of the grey partridge and hypothesized that (1) the populations of the species in Europe, and especially in northern marginal populations in Finland, show genetic structuring, resulting from inhibited gene flow between populations (2) the post-glacial colonization of Finland occurred from the eastern refugia, in contrast to the rest of Europe, which was colonized mainly from the southern refugia, and (3) the farm stocks used for introductions differ from the wild populations in their genetic structure.

Materials and methods

Sampling of birds and laboratory methods

We sampled 227 grey partridges (159 wild and 68 farm birds) for this study (Table 1, Fig. 1). We extracted DNA from blood and other tissues using the methods described by Kvist et al. (1998) and Sambrook et al. (1989), or the PureGene DNA extraction procedure (GENTRA Systems Inc., Minneapolis, MN, USA), according to the manufacturer's instructions.

Table 1.  The origin and number of the grey partridges ( Perdix perdix ) used in this study.
CountryCodePlaceWild/captiven
AustriaAUSNorth from ViennaWild2
BulgariaBULBreznik, ChirpanWild6
EnglandEN1DurhamWild8
EN2NorfolkWild8
EstoniaEST – /South from TallinnMuseum/wild3 (1/2)
FinlandFI1Oulu-regionWild9
FI2South OstrobothniaWild26
FI3JokioinenWild6
FI4South FinlandWild3
FFISouth OstrobothniaCaptive16
FranceFR1AisneWild6
FR2BeauceWild5
FR3PyreneesWild12
GermanyGE1Bissendorf, HannoverWild18
GE2Feuchtwangen, BavariaWild5
GreeceGREGrammoeWild10
HungaryFHUSopronCaptive15
IrelandIREBooraWild8
ItalyITAUdineWild5
 FITTuscanyCaptive26
KazakhstanKAZKustanai, BozhakoliWild2
LatviaLATWild1
PolandPOLCzempin, LublinWild14
RussiaRUSWild1
SpainFSPNorth SpainCaptive3
SwedenSWEMellby, Ottenby, on ÖlandWild4
FSWTranås, mainlandCaptive8
Total n  Wild/captive227 (159/68)

We extracted the DNA from feather quills using the method described by Kvist et al. (1999). The top of the feather quill (3–5 mm) was cut into a tube and covered with 100 μL of sterile water. Samples were then incubated at 37 °C for 45 min with 2 μL of proteinase K (20 mg mL−1), after which 30 μL of 20% Chelex-100, 4.6 μL of 1 m DTT and 2 μL of proteinase K (20 mg mL−1) were added. Samples were then incubated for 40 min. Afterwards the samples were briefly vortexed and centrifuged, and then boiled for 7 min. Samples were stored at +4 °C for PCR.

For the PCR of the whole CR, the forward primer L16620 (5′ACCCCATAATACGGCGAAGGATT) and the reverse primer H1273 (5′TCTTGGCATCTTCAGTGCCATGCTTT) were designed. These primers were based on the conserved regions located by comparing the chicken Gallus gallus (Desjardins & Morais, 1990) and the quail Coturnix coturnix (Desjardins & Morais, 1991) tRNAGlu and tRNAPhe genes available in GeneBank (accession numbers X52392, and X57245, respectively). After successful amplification and sequencing of the whole CR, more specific primers were designed. For population analysis, we amplified the first 410 nucleotides of the CR (downstream from the tRNAGlu), i.e. the control region 1 (CR1), which is the most variable of the three domains of the CR in the grey partridge. CR1 was amplified with the forward primer LPPGLU (5′CACTGTTGTTCTCAACTACAGG), and the reverse primer H414 (5′GGTGTAGGGGGAAAGAATGGG).

Amplification took place in 50 μL (containing 2–5 μL of DNA) using DyNazyme™ II DNA Polymerase (Finnzymes) for one cycle in 94 °C for 2 min, followed by 33 cycles in 94 °C for 1 min, 59 °C for 0.5 min and 72 °C for 1 min. The final extension at the end of the profile was 72 °C for 5 min. Negative controls were carried along the PCR reactions to detect contamination. Amplified PCR product was purified from 1% agarose gel using polyester plug spin inserts (Glenn & Glenn, 1994).

Sequencing was carried out using the ABI Prism Dye Terminator cycle sequencing ready reaction kit (Perkin–Elmer) for ABI 377 DNA sequencer following the recommendations of the manufacturer. PCR primers were used as sequencing primers to detect CR1. Sequencing primers for the whole CR were L156 (5′CCATCTCTCCTTACGTG), L438 (5′TCACGTGAAATCAGCAACCC) and L717 (5′TGCGGAGTACTACTCAA), and for the CR1 LPPGLU (5′CACTGTTGTTCTCAACTACAGG).

Sequence comparisons and statistical analysis

Sequence alignment was done by eye in SEQLAB (Wisconsin Package Ver. 9.1). The whole CR was sequenced from 14 birds. The first 20 nucleotides were ignored because of incompleteness in sequencing, although they included one informative site at position 7 (transition T/C, not included in the Appendix). The minimum-spanning network (Fig. 2) was constructed manually based on the segregating sites (Appendix) and Arlequin 2.0. The two main sequences of P. perdix were deposited in GeneBank with the accession numbers AF115404 (main eastern haplotype) and AF115405 (main western haplotype).

Figure 2.

The minimum-spanning network of the CR1 haplotypes of the grey partridge ( Perdix perdix ) drawn based on segregating sites. Number of individuals of given haplotype is inside parentheses. Haplotype name is given outside the circles. Each line between two crossbars or circles represents one mutation.

To track diversity in the populations, basic diversity estimates were calculated. Estimates for haplotype diversity (h) and nucleotide diversity (π) were calculated from the formulas h = (n/n−1)(1-Σinline image), and π = (n/n−1)Σxixjπij (Nei, 1987). In these formulas xi = population frequency of the ith haplotype; xj = population frequency of the jth haplotype, and πij = proportion of nucleotide differences between ith and jth haplotypes. The number of polymorphic sites per nucleotide site (θs) was estimated. Parameters h, π, and θs were calculated using DnaSP 3.51 (Rozas & Rozas, 1997). MEGA (Kumar et al., 1993) was used to compute Jukes–Cantor pairwise genetic distances.

The mutation rate in the CR is variable among taxa. To approximate the coalescence times for main haplotype clades the conventional mutation rate of 2%/Myr for the whole CR was used (equation 2 in Avise et al., 1988). This rate is used for the Old-world partridges and fowl (Galliformes), and for the genus Alectoris (Randi, 1996), which supports the use of this mutation rate estimate for the grey partridge.

Population differentiation was assessed using the φST statistics of amova 1.55 (Arlequin 2.0, Excoffier et al., 1992). Specimens within the range of 300 km were pooled. Population samples of at least five individuals were included in the analyses. amova was conducted between clades (eastern–western), among all populations (BUL, EN1, EN2, FI1, FI2, FI3, FR1, FR2, FR3, GE1, GE2, GRE, IRE, ITA, POL), and separately for both clades including only ‘pure’ populations. A ‘pure’ population indicates populations which contain either western or eastern haplotypes, but not both.

Matrilinear gene flow (Nmf) was estimated from φST = 1/(2Nm + 1) (Nei, 1987). The relationship between population differentiation and geographical distance was examined using the Mantel's test. As an alternative gene flow estimate, we used a maximum-likelihood (ML) method based on the coalescent theory (Beerli & Felsenstein, 1999) using the program MIGRATE (version 0.9.7; Beerli, 1997). MIGRATE estimates migration rate (expressed as the parameter Nfm) from a sample of every possible genealogy of sampled sequences. It takes into account the history of mutations and the uncertainty of the genealogy. MIGRATE allows independent estimates of unequal migration rates and different population sizes, in contrast to FST-based estimates. Great Circle Distance was used to calculate the distances between separate populations for the gene flow estimations.

To reveal a possible post-glacial colonization route from Pyrenees northwards we analysed the existence of the south–north diversity cline within the western clade. If such cline existed, there should be significantly negative correspondence between population genetic diversities (π and h) and the population location (latitude). This was tested using the linear regression analysis (SPSS 8.0).

Expressed mtDNA variation is assumed to be almost independent of environmental selection pressures (but see Ballard & Kreitman, 1995; Fry, 1999). Changes in population size may result in deviations from the neutral patterns of nucleotide variation expected at equilibrium. Using the frequency distribution of mutations segregating in extant populations, the magnitude of a deviation can be measured by Tajima's (1989)D-statistic, which was calculated using DnaSP 3.51.

The expected distributions of the pairwise genetic distances under models of population expansion or equilibrium (Rogers & Harpending, 1992; Rogers, 1995) were calculated using DnaSP 3.51. Here τ describes time in units of ½u generations, where u is the sum of per-nucleotide mutation rate in the DNA region under study. θ is the expected pairwise difference that increases from θ0 to θ1 at τ units of mutational time before present. To find the possible equivalence between expected and observed distributions, they were compared with the χ2 test (SPSS 8.0).

Results

Sequence variation

The length of the whole mitochondrial CR of the grey partridge was 1151 nucleotides (sequenced from 14 individuals). Two main mitochondrial DNA haplotype clades, which differed from each other by 25 nucleotides (2.2%), were detected. Along the 390 bp of the 5′ end of the control region (CR1), the difference was 14 nucleotides (3.6%) between the two basic haplotypes in these clades. With reference to the geographical origin of the clade haplotypes, they are referred to as ‘eastern’ and ‘western’. Both the western and the eastern clade included one basal haplotype, which is ancestral to several other haplotypes (Fig. 2). The basal western haplotype (MW) was observed in 115 individuals, and the basal eastern haplotype (ME) in 18 individuals.

The readable 390 nucleotides of the CR1 sequences from the sampled 227 grey partridge exhibited 49 (12.6%) variable sites, of which 24 (49.0%) were phylogenetically informative. One insertion/deletion (1 bp), 11 transitions and two transversions were found, and 45 CR1 haplotypes were identified (Fig. 2, Appendix). The mean nucleotide composition of the species in the CR1 – A 27.0%, T 32.4%, C 26.4% and G 14.1%– was similar to several other bird species (Lucchini & Randi, 1998; Kvist et al., 1999; Holder et al., 2000).

Basic diversity parameters of the populations are presented in Table 2. Haplotype diversity (h) for the wild birds in the eastern populations varied between 0.410 ± 0.103 (SD) in FI2, and 0.800 ± 0.122 in FI3, and for the western populations from 0.000 (GE1) to 0.900 ± 0.161 (GE2). The mean nucleotide diversity (π) for the wild birds in the eastern lineage was 0.00548 ± 0.00095 (SD) and in the western lineage 0.00317 ± 0.00068. The maximum nucleotide diversity was 0.01902 ± 0.00550 in populations BUL and IRE, which contained both western and eastern haplotypes. The GE1 population was monomorphic. θs values were in concordance with the π values (Table 2).

Table 2.  The sample size ( n ), nucleotide ( π ) and haplotype ( h ) diversities, θ estimated from the number of polymorphic sites per nucleotide ( θ s), Tajima's D and its significance * ( P ) for the analysed grey partridge ( Perdix perdix ) populations.
Populationsn π SDhSD θ s SDTajima's DP
Wild
FI190.017570.002600.750150.1120.012300.005962.04494P  < 0.05 *
FI2240.001430.000520.409500.1030.002060.00131 −0.76831P  > 0.10
FI360.002740.000570.800000.1220.002250.001751.03194P  > 0.10
EN180.002580.001860.250000.1800.003980.00248 −1.534700.10 > P > 0.05
EN280.005140.003710.250000.1800.007930.00424 −1.70118P  < 0.05 *
IRE80.019020.005540.700000.2180.017280.009470.73751P  > 0.10
FR160.001370.000440.533000.1720.001130.001130.85057P  > 0.10
FR250.003600.001120.700000.2180.003700.00261 −0.17475P  > 0.10
FR3120.004480.001520.682000.1480.007660.00376 −1.685900.10 > P > 0.05
GE1180.000000.000000.000000.0000.000000.00000 
GE250.003610.000980.900000.1610.003710.00214    0.17475P  > 0.10
GRE100.001420.000520.511100.1640.001810.00138 −0.69098P  > 0.10
BUL60.019020.005500.800000.1720.016890.008800.77938P  > 0.10
ITA50.009250.002700.600000.1750.007400.004521.71830P  > 0.10
POL140.000740.000420.275000.1480.001620.00122 −1.48074P  > 0.10
Farm
FFI160.000960.000440.350200.1480.002320.00150  
FSW80.009460.005960.607300.1640.013880.00683  
FIT260.004690.001840.576000.1630.006830.00343  
FHU150.000690.000390.257200.1420.001580.00119  
FSP30.001710.00081  0.001710.00171  

All but one farm stock birds from Finland, Hungary, Italy and Sweden represented western haplotypes. The exceptional eastern haplotype (FSW, E15), was found in the Swedish farm. In farm birds, π and h varied mostly between the Hungarian and Swedish farm birds (π from 0.00069 ± 0.00039 to 0.00946 ± 0.00596, and h from 0.25720 ± 0.142 to 0.60730 ± 0.164, respectively).

The maximum Jukes–Cantor pairwise genetic distance was 0.0565 (between W17 from EN1 and E13 from BUL). The maximum distance inside the western lineage was 0.0345 (between W19 from EN1 and W21 from a Finnish farm, FFI) and inside the eastern lineage 0.0104 (E9 from Sweden to E14 from BUL and E6 from FI4, as well as between E14 from BUL and E6 from FI4).

Phylogeographical patterns

The two basal haplotypes, ME and MW, were geographically widespread. ME was found in Bulgaria (BUL), Finland (FI1), Greece (GRE) and Ireland (IRE). MW was found in Austria, England (EN1, EN2), France (FR1, FR2, FR3), Germany (GE1, GE2), Poland (POL) and Sweden. The more rare haplotypes were consequently found at one locality each. One western haplotype (W17) was found in the wild in Finland (FI1) and Germany (GE2), in a Hungarian (FHU) and an Italian (FIT) farm, and in a museum specimen in Latvia (LAT). The wild population in Italy (ITA) included two unique haplotypes, and the Pyrenean (FR3) population one main type and five different haplotypes related to the French types. In addition to the ME haplotype, two other eastern haplotypes (E2, E3) were also well represented in Finland. Further, eastern haplotypes were found in Estonia, Kazakhstan, Russia, and on the Island of Öland, Sweden.

We could not find any south–north cline in the western clade either in the nucleotide (r = 0.009, P = 0.795) or in the haplotype diversity (r = 0.087, P = 0.409). If the populations EN1 and EN2 with unique haplotypes (W18, W19) were removed from the analysis, a slight trend of decreasing diversity with increasing latitude could be seen.

Population structure, gene flow and changes in the population size

All amova results are presented in Table 3. Of the observed variation, 88.4% was explained by the differentiation of the western and eastern clades. Independently, variation among populations explained 78.1% of the variation, most of this resulting from the geographical separation of the clades. When only ‘pure’ populations were included in the clades, the among-populations variance explained 24.5% of the variation in the western clade, and 55.1% in the eastern clade.

Table 3.  Results of amova based on four different categorizations. ‘Pure’ populations are populations including only haplotypes of either of the clades.
amovaSource of variationd.f.% of variation
1. Clades (eastern–western)Among    188.4
Within12111.6
2. All populationsAmong  1478.1
3a. Western clade ‘pure’ populationsAmong    824.5
(EN1, EN2, FR1, FR2, FR3, GE1,  GE2, ITA, POL)Within  7275.5
3b. Eastern clade ‘pure’ populationsAmong    255.1
(FI2, FI3, GRE)Within  3744.9

Several significant φST values and the low estimated number of migrants (Table 4) between populations showed that most populations were differentiated from each other. In most cases, the estimated number of migrants between populations was less than one individual per generation (Table 4). The ML estimates obtained from MIGRATE were consistently one of ten of the estimates based on φST values (data not shown), most probably depending on the different basis of these two estimates. The Irish population did not differ from geographically distant populations FI1, GRE and BUL. The negative association between the geographical distance and the gene flow estimates between populations was revealed by Mantel's test (n = 15, g = −1.67, r = −0.68, P = 0.005).

Table 4.  Pairwise φ ST s between populations of the grey partridge ( Perdix perdix ) above diagonal, and gene flow estimates (number of female migrants per generation Nm) below diagonal.
PopulationsFI1FI2FI3EN1EN2IRLFR1FR2FR3GE1GE2GREBULITAPOL
  1. Significant values (P < 0.001) are in boldface.

FI1 0.637240.409120.416630.43733 −0.125040.423130.386800.515400.619790.359590.37386 −0.112390.395610.54274
FI20.3 0.091020.941090.955620.670690.963120.960000.946990.977870.956930.683360.623080.946080.96227
FI30.74.9 0.892670.929180.392150.944960.934350.915730.982780.926380.600750.349430.891140.95111
EN10.70.00.1   0.001180.541270.02379 −0.031750.25447  0.110330.035280.910200.533980.320980.03722
EN20.60.00.0424.6 0.57523 −0.03554 −0.066950.29216  0.110330.091470.941060.566690.411660.01728
IREinf.0.20.80.4  0.4 0.567800.524300.640970.782200.495940.32843 −0.175030.495750.69548
FR10.70.00.020.5inf.0.4  −0.170970.313270.446150.192060.957490.551620.446510.10371
FR20.80.00.0inf.inf.0.51inf. 0.275720.284290.100780.951120.512410.384310.02478
FR30.50.00.01.5  1.20.31.1    1.3 0.504860.323380.924960.628320.231910.36567
GE10.30.00.04.0  4.00.10.61.30.5 0.479870.984580.756650.720150.01821
GE20.90.00.013.7  5.00.52.1    4.51.0  0.5 0.943500.483030.392940.16175
GRE0.80.20.30.0  0.01.00.0    0.00.0  0.00.0 0.262650.915320.95591
BULinf.0.30.90.4  0.4inf.0.40.50.3  0.20.51.4 0.488250.67664
ITA0.80.00.11.1  0.70.50.60.81.7  0.20.80.00.5 0.52865
POL0.40.00.012.928.40.24.319.70.927.02.60.00.40.2 

The significantly negative Tajima's D in both the western (D = − 2.23189, P < 0.01) and the eastern clade (D = −1.83174, P < 0.05) is characteristic for an expanding population after a bottleneck. In the western clade, the observed distribution of pairwise genetic differences (τ = 0, θ0 = 1.52658 and θ1 = 500), which did not follow the model of either equilibrium or expansion from a single bottleneck (χ2 = 37.4462, P = 0.001, Fig. 3), and some unique haplotypes may indicate some geographical structuring. In the eastern clade, the observed distribution did not deviate (χ2 = 8.727, P = 0.190) from the model of population expansion from one single bottleneck (τ = 1.56, θ0 = 0, and θ1 = 500, Fig. 3). However, geographical structuring could be seen in the minimum-spanning networks of both clades.

Figure 3.

Expected and observed distributions of the pairwise nucleotide differences from the CR1 of the grey partridge ( Perdix perdix ) representing western and eastern clades.

The mutation rate of 2%/Myr for the whole CR lead to an estimated divergence time of c. 1.1 Myr between the clades. The coalescence time for the eastern clade was estimated to be c. 440 000 years based on the mean pairwise genetic distance of 0.008886 and for the western clade c. 330 000 years based on the mean pairwise genetic distance of 0.006576.

Discussion

Refugia and recolonization history of the grey partridge

Our hypothesis on separate eastern and southern refugias of the grey partridge was supported by the minimum-spanning network, which reflected the ancient history of the species. There was one basic haplotype in both the western and the eastern clade. This suggests that there were at least two separate refugia where ancestral P. perdix survived the glaciations. After the last glaciation, the northern regions of Europe were mainly recolonized from Iberian and Balkan refugia with the Alpine barrier isolating the Italian lineage (Taberlet et al., 1998). The western mtDNA haplotypes were found primarily in England, France, Germany, Austria, and Poland, indicating an origin in Iberian refugia. The divergent haplotype found in Italy (W3) among wild birds, may reveal a separate Italian refugium. In addition, one unique haplotype was found in northern England. The origin of this bird is unknown.

The two potential refugia for the eastern lineage are the Balkan and the Caucasian refugia (Taberlet et al., 1998). As our samples from Kazakhstan and Russia represented the eastern mtDNA lineage together with the Bulgarian, Finnish and Greek samples, we assume that the eastern lineage expanded from either of these refugia. The distribution pattern of the eastern haplotypes suggests a possible post-glacial recolonization route of the Finnish grey partridge from Caucasia across Russia.

The Irish wild grey partridge population was similar to the Greek population. Modern introductions have been conducted with birds from Central Europe, and the two birds representing western haplotypes may be the remnants of these releases. However, the eastern haplotypes may have an older origin, and man-aided colonization may have followed the same route as described for the house mouse M. musculus by Hewitt (1999), from Caucasia via the Mediterranean to Ireland.

In this work, significantly negative Tajima's Ds, which could indicate expanding populations, were found in both clades. In the western clade, the model of an expanding population after one single bottleneck was rejected. The western clade showed a haplotype network with one main haplotype, several very close haplotypes and also some quite distinct haplotypes. Observed deviations from the neutral patterns of nucleotide variation expected at equilibrium may, in general, result from past changes in the population size (Aris-Brosou & Excoffier, 1996; Fry & Zink, 1998). The existence of the distant haplotypes (an additional peak in the curve of observed pairwise genetic differences, see Fig. 3) may indicate remains of ancient (i.e. pre-bottleneck) structuring among western birds. Since τ was 0, this may reveal a very recent bottleneck (Rogers & Harpending, 1992), possibly dating back to the population crash at the 1950s.

In the eastern clade, the expansion of the population was supported by the significantly negative Tajima's D, and the distribution of the pairwise genetic differences. However, the form of the minimum-spanning network was also reflecting structuring, which may result from random sorting of haplotypes, amplified by sampling bias. The eastern E3 haplotype was over-represented as a result of a relatively dense sampling in FI2 population, where this haplotype was common (20 of 22 individuals).

The divergence (3.6%) between the eastern and western mtDNA clades in the CR1 was of the same magnitude than between Sicilian and Alpine haplotypes (3.7%) of the rock partridge Alectoris graeca (Lucchini & Randi, 1998). With the conventional 2%/Myr mutation rate the divergence time between the mtDNA lineages dated back to early Pleistocene (1.1 Myr). This timing is earlier than the average phylogenetic subdivision times for several avian species (Shields & Wilson, 1987; Wenink et al., 1993; Avise & Walker, 1998; Lucchini & Randi, 1998).

Population structuring and the estimation of genetic variation within and among populations

The results of amova supported our hypothesis of strong genetic structuring. Most of the genetic structure could be explained by the differentiation between the clades, and their vicariant geographical distribution. The population structure found in the European grey partridge was relatively strong in comparison to the blue tit Parus caeruleus (Kvist et al., 1999), the dunlin Calidris alpina (Wenink et al., 1996) and the greenfinch Carduelis chloris (Meriläet al., 1997). However, the pairwise φST values were similar to those of the Mediterranean populations of the rock partridge (Lucchini & Randi, 1998). Geographical segregation of haplotypes reflects allopatric divergence of refugial populations and subsequent low maternal gene flow among the populations. Both gene flow estimates support the hypothesis of extremely low gene flow between populations. The grey partridge is relatively sedentary, and the short natal dispersal combined with the high site-fidelity of adults may facilitate geographical structuring. The mean observed dispersal distance of released grey partridges is 5.96 ± 1.8 km (Finnish Game and Fisheries Research Institute; Ringing Centre, Finnish Museum of Natural History). The breeding dispersal distance of radio-tagged wild hens is on average 3.1 ± 0.5 km and that of released hens 2.32 ± 0.8 km (Putaala & Hissa, 1998). Mantel's test revealed a negative correlation between the geographical distance and the genetic differences between populations. This result can be explained by the genetic similarity of distant populations BUL, GRE, FI1, FI2, FI3 and IRE, and the similarity of the eastern clade haplotypes in them.

Nucleotide diversity varied substantially among populations. Grassland habitat destruction, which has been intensified by severe hunting pressure, has occurred in the areas of low diversity, and for example GRE has been bottlenecked during the last century. Only one haplotype was found in a large area in Germany (GE1), which may result from a local extinction, followed by random drift during recolonization. Unusually high nucleotide and haplotype diversities were recorded in populations with a mixture of western and eastern haplotypes (FI1, IRE, BUL).

Introductions – aims in hunting or conservation?

In Finland, grey partridges are reared in captivity and released into the wild, with the aim to strengthen natural populations. Survival of released grey partridges has been poor in the wild (Putaala & Hissa, 1998). Both Finnish and Swedish farm birds represented the western mtDNA clade (with one exception, an imported wild bird from population FI2 to the farm in Sweden, V. Mikkilä, pers. comm.), whereas the wild population in Finland represented the eastern clade. Despite the frequent releasing of birds representing the western clade in the FI2 population, no birds of this clade were found in the wild in this area. In FI1 population, five grey partridges representing the western clade were found, as remains from very recent releases for research purposes (Putaala & Hissa, 1998). The Italian farm stock consisted mainly of the basal western haplotype, and the wild birds represented two different unique haplotypes. Native Italian P. p. italica was supposed to still exist back in 1975 (Lovari, 1975), but unfortunately, frequent releasing of grey partridges has occurred in Italy, and the Italian subspecies in its purest form is presumed extinct (Matteucci, 1988).

Introductions may fail because of maladaptation of birds to the environmental conditions of the introduction areas. Birds may be released for hunting purposes only, without any target to increase population size. Some of the released birds may, however, survive in the wild and breed later. If birds are hand-reared for conservation purposes, the genetic origin of these birds should be compatible with the wild birds of the releasing area (Rave et al., 1994; Glenn et al., 1999). However, habitat protection, as well as restrictions of hunting, may be essential to preserve the grey partridge populations.

Acknowledgements

We are grateful to following people who kindly provided us with samples: J. Alhainen, L. Ahlström, J. Bisi, P. Haapala, J. Heikkilä, A. Heikka, S. Hietikko, T. Hill, J. Kalmari, E. Knuuttila, J. Kojonen, M. Leinonen, K. Luhtala, H. Merviö, J. Mäkelä, A. Putaala, J. Rinne, M. Syrjälä, S. Takamaa, J. Tiainen, A. Turtola, B. Vuorela, T. Loukomies, R. Westerinen, A. Wahlberg (Finland), C. Ramel (Sweden), S. Stoyanov (Bulgaria), S. Effenberger (Austria), B. Kavanagh & C. O'Gorman (Ireland), R. Draycott, D. Potts, A. Putaala & M. Swan (UK), W. Kaiser & E. Strauss (Germany), J. Nadal Garcia (Spain), F. Reitz, E. Bro, C. Novoa (France), B. Alexiou (Greece), P. Tout (Italy), M. Panek (Poland), The Lesser White-Fronted Goose-expedition group (Kazakhstan), R. Ilisson (Zoological Museum, University of Tartu, sample An 5254) & M. Leinonen (Estonia), S. Faragó (Hungary), E. Lukševiš (Latvia). The Finnish Game and Fisheries Research Institute and the Ringing Centre of the Finnish Museum of Natural History allowed us access to the ringing data.

We thank Prof. P. Pamilo, Dr L. Kvist, Dr M. Ruokonen, and three anonymous reviewers for valuable comments on this paper. Dr D. Potts, Prof. R. Gutiérrez and R. Thomson kindly revised the language. H. Parkkinen, S. Finne and E. Jauhiainen helped in the laboratory.

This study was financed by Alfred Kordelin Foundation, Emil Aaltonen Foundation, Finnish Cultural Foundation, Finnish Ministry of Agriculture and Forestry, South Ostrobothian Fund for Finnish Cultural Foundation, University of Oulu, University Pharmacy Fund and the Thule Institute at the University of Oulu.

Table Appendix. 

The variable sites in the grey partridge ( Perdix perdix ) control region 1 (CR1). First nucleotide in the CR is number one.
  1. First 20 nucleotides are excluded from the analysis due to sequencing ambiguities. ME = main eastern haplotype, MW = main western haplotype.

 111111122222222222222333333333333333334444
 777999017888911122344557899000011345568899990000
 467135515689201703138798914028908891201634790245
METAAGGTTTACTTGCTTTTCTAAATATCTACTCTGTATCCCTACATATA
E2 .......... .......... .......C.. .......... ........
E3 .......... .......... .......C.. .........T ........
E4 .......... .......... .......C.. ........AT ........
E5 .......... ......C… .......... .......... ........
E6 .......... .......... .......... .........T..T.....
E7 .......... ......CG.. .......... .......... ........
E8 .......... ............G….C.. .........T ........
E9 .......... ............G..C…. .......... ........
E10 .......... .......... …..C…. .......... ........
E11 .......... ..........G......... .......... ........
E12 .......... .........C .......... .......... ........
E13 .......... .......... ............G.G..... ........
E14 .......... .......... .............T...... ........
E15 ......C... .......... .......... .......... ........
MWCG-.....G..C....A..C..G…T.C.CT..........T.....
W2CG-C….G.......A..C..G…T.C.CT..........T.....
W3CG-..A..G..…C.G..C..G…T.CTCT..........T.....
W4CG-.A…G..C….A..C..G…T.C.CT..........T.....
W5CG-.A…G..C….A..C..G…T.C.CT.......-..T.....
W6CG-.A…G..C….A..C..G…T.C.CT.......T..T.....
W7CG-.A…G..C…CA..C..G…T.C.CT..........T.....
W8CG-..…G..CT…A..C..G…T.C.CT..........T.....
W9CG-…..G..C..C.A..C..G…T.C.CT..........T.....
W10CG-…..G..C..C.A..C..G…T.C.CT..........TTA...
W11CG-…..G..C….A..C..G…T.C.CT........G.T.....
W12CG-….CG..C….A..C..G…T.C.CT.....T....T.....
W13CG-…..G..C….AC.C..G…T.C.CT..........T.....
W14CG-…..G..C….A…..G…T.C.CT..........T.....
W15CG-…..G..C….A…..GG..T.C.CT..........T.....
W16CG-…..G..C….A..C..G...T.C.CT.T........T.....
W17CG-…..G..C….A..C..G…..C.CT..........T.....
W18CG-.A…GTCC….A.-C..G…T.C.CT..........T.....
W19CG-…..G..C….A..C..G.T.GACTCT.........CT..TAT
W20.G-..…GT ........TC..G…T.C.CT..........T.....
W21CG-…..G. ......G.-C..G…T.C.CT…TA.A...T.....
W22CG-…..G..C….A.-C..G…T.C.CT..........T.....
W23CG-…..GT.C….A..C..G…T.C.CT..........T.....
W24CG-…..GT ......A..C..G…T.C.CT..........T.....
W25CG-…..GT …T..A..C..G…T.C.CT.T........T.....
W26CG-…..G..C..C.A..C..G…T.C.CT..........T.....
W27CG-…..G..C….A..C.GG…T.C.CT..........T.....
W28CG-…..G. ......A..C..G…T.C.CT..........T.....
W29CG-…..G..C….A..C..G…T.C.CT…..T....T.....
W30CG-…..G..C….A..C..G…T.C...T.........T.....

Ancillary