We studied the genetic structure of a hybrid zone between Red-legged (Alectoris rufa) and Rock (A. graeca) partridges in the southern French Alps using six allozyme loci. Allele frequencies showed sharp clinal variation across the hybrid zone, shifting by 60% over a distance of about 60 km, on average. Single-locus clines were coincident, but only partially concordant, with intermediate allele frequencies in the hybrid population. Percent polymorphic loci, heterozygosity, and linkage disequilibria were higher in the hybrids and in some populations near the hybrid zone than in allopatric populations. Nonrandom associations favored parental allele combinations. The hybrids and most of the allopatric populations were in Hardy-Weinberg equilibrium, but six populations near the hybrid zone showed a significant deficit of heterozygotes. The partridges in the hybrid zone included F 1 hybrids, backcrosses, and other recombinant genotypes. This hybrid zone could result from secondary contact of formerly allopatric populations following deglaciation of the Alps, probably not before 6,000 to 8,000 years ago. The width of the observed multilocus cline (70 to 160 km) is much shorter than expected (1,120 to 2,750 km) by neutral diffusion of allelic variants since the time of secondary contact of the two species (2,000 to 3,000 generations). Linkage disequilibria suggest that this hybrid zone is maintained by crossing of genetically divergent parentals and natural selection against hybrid genotypes. Although parapatric interspecific populations can exchange their genes through the hybrid zone, natural selection might constrain gene flow such that two species continue to evolve independently. Received 12 February 1998, accepted 23 July 1998.

Istituto Nazionale per la Fauna Selvatica, Via Cg Fornacetta 9, 1-40064 Ozzano dell'Emilia (BO), Italy; and 20ffice National de la Chasse, CNERA-Faune de Montagne, B.P. 38, F-06201 Nice Cedex 3, France THE GENETIC STRUCTURE of inter- and intra- specific hybrid zones is usually characterized by abrupt and concordant clinal variation in al- lele frequencies between the parental popula- tions (Barton and Hewitt 1985, 1989; Harrison 1990, 1993; Hare and Avise 1996). Mosaic hy- brid zones (Harrison 1986) can show complex clines owing to differential habitat preferences of the different hybrid genotypes (Rand and Harrison 1989, Howard and Waring 1991, Sites et al. 1995). Hybrid zones appear to be some- what stable in time and space (Barton and Hewitt 1985, 1989) and could be maintained by dynamic equilibria between gene flow via dis- persal and natural selection against hybrid ge- notypes (Barton and Hewitt 1989). Alternative- ly, transient hybridization, which results from range expansion and recent contact between reproductively compatible taxa, produces mov- ing fronts of introgression that eventually lead to local genetic extinction of the introgressed populations (Gill 1994, 1997). Hybrids can be intrinsically unfit and then eliminated by selec- E-mail: met0217@iperbole.bologna.it tion due to endogenous factors ("tension zones"; Key 1968, Barton and Hewitt 1985, 1989; Searle 1993, Sites et al. 1995). Alterna- tively or concomitantly, hybrid genotypes can be exposed to exogenous selection pressures in different habitats ("clinal" models; May et al. 1975, Endler 1977) and occasionally can be more fit than parentals at particular transition- al locales ("bounded hybrid superiority" mod- el; Moore 1987, Arnold and Hodges 1995, Emms and Arnold 1997). Hybridization is relatively common in birds (Grant and Grant 1992), and many avian par- apatric distributions have been described (e.g. Prigogine 1980, Rising 1983) that apparently represent stable zones of overlap and hybrid- ization (sensu Short 1969). Hybridizing avian populations often show low geographic varia- tion and absence of diagnostic alleles at both nuclear and mitochondrial loci (Barrowclough 1980, Brown and Robbins 1986, Grudzien and Moore 1987, Corbin and Wilkie 1988, Moore et al. 1991, Saino et al. 1992). Morphological traits usually have sharper clinal transitions, which could be controlled by natural and sexual se- lection (Moore 1987, Gill 1997). Some avian hy- brid zones appear to be stable (Moore and Buchanan 1985), whereas others are transient (Gill 1997) or have shifted their geographic po- sition during historic times (Cook 1975, Rising 1983). The Red-legged Partridge (Alectoris rufa; dis- tributed from Iberia to France and northwest- ern Italy) and Rock Partridge (A. graeca; dis- tributed throughout the Alps, central and southern Apennines and Sicily in Italy, and the Balkans, reaching Albania and Greece; Johns- gard 1988) are largely allopatric but hybridize along the border of the southern French Alps (Bernard-Laurent 1984; Fig. 1). As a conse- quence of climate deterioration at the Pliocene/ Pleistocene boundary, the Red-legged and Rock partridges diverged probably in western and eastern refugia, respectively, and then ex- panded during the Holocene postglacial warm- ing to a secondary contact zone (Randi et al. 1992, Randi 1996). These two species are ge- netically well differentiated and probably are sister taxa (Randi 1996). They have diagnostic plumages that allow identification of pure and hybrid birds (Table 1). The morphologically hy- brid population presently is distributed over a band of open habitat, about 15 km wide, at moderate elevations along the southern edge of the French Alps (Bernard-Laurent 1984; Fig. 1). In the past, however, the hybrid zone was prob- ably wider, ranging from the northern French Alps to the western Apennines and the Ligur- ian Alps in Italy (Degland and Gerbe 1867, Spantb 1979). To date, the genetics of this partridge hybrid zone has not been studied. Here, we report on the genetic structure of the hybrid population relative to Red-legged Partridge and Rock Par- tridge populations close to and far from the hy- brid zone. Results of a survey of six enzyme loci are used to describe single- and multilocus al- lelic clines and introgression across the hybrid zone, as well as patterns of allele diversity and linkage relationships between loci. Genetic re- sults are integrated with palaeoclimatic, eco- logical, and behavioral data to suggest the pos- sible origin and dynamics of this hybrid zone. MATERIALS AND METHODS During the 1989 to 1993 hunting seasons, we col- lected 95 Red-legged Partridges and 320 Rock Par- tridges (hereafter called the "allopatric popula- tions") near or far from their contact zone (Fig. 1). Fourteen allopatric populations were sampled from the southern French and Italian Alps to the eastern Alps. Four other allopatric populations were sam- pled outside the Alps: two western Red-legged Par- tridge populations from Portugal and Spain, and two eastern Rock Partridge populations from central Ap- ennines (Italy) and Albania. We collected 38 par- tridges within the area of species' contact and over- lap (the "hybrid zone" across the southern French Alps). The plumage patterns of these birds were an- alyzed using the criteria of Bernard-Laurent (1984); 30 of the birds were morphological hybrids that ex- hibited various combinations of species-specific plumage traits (population no. 4; the "hybrid popu- lation;" see Fig. 1, Table 2), 7 had typical Rock Par- tridge plumage (population no. 5; "Rock Partridges within the hybrid zone"), and 1 was a pure Red-leg- ged Partridge (excluded from the following genetic analyses). All birds collected outside the hybrid zone exhibited morphological traits of pure Red-legged or Rock partridges, depending on sampling locations. Liver and heart tissue samples were collected and stored frozen at -80C. About 0.5 g of each tissue sample was separately homogenized in 1-mL 0.01 M Tris/HC1 pH 7.5, 1 mM Na 2 EDTA, and 1 mM [3-mer- captoethanol buffer and centrifuged for 15 min at 17,000 rpm at 4 C. Clear supernatants, diluted in 1 volume of 40% glycerol, were aliquoted in microtiter plates. Polyacrylamide gel electrophoresis was used to study allelic variability of the following six loci, which had high intrapopulation polymorphism and differed in allele frequencies between the two species of partridges (Randi et al. 1992): (1) liver ct-naphthyl- acetate esterase-2 (Est; E.C. No. 3.1.1.1); (2) cytoplas- mic glutamate-oxaloacetate transaminase (Got; E.C. No. 2.6.1.1); (3) cytoplasmic malic enzyme (Me; E.C. No. 1.1.1.40); (4) cytoplasmic isocitrate dehydroge- nase (Idh; E.C. No. 1.1.1.42); (5) mannose phosphate isomerase (Mpi; E.C. No. 5.3.1.8); and (6) amylase (Amy; E.C. No. 3.2.1.1). Electrophoretic conditions are given in Randi et al. (1992). Electromorphs were treated as alleles and coded by their mobility from the origin, with the most anodal allele named "A," and so forth. Because intrapopulation allele frequencies did not fluctuate significantly among sampling years (Work- man and Niswander's contingency test; data not shown), we pooled all samples collected at the same locality in different years. The electrophoretic data were analyzed using the following software and sta- tistical methods: (1) BIOSYS-1 (Swofford and Selan- der, 1989) was used to compute allele frequencies, gene-diversity estimates (P = percent polymorphic loci; A = mean number of alleles per locus; Ho and H c = observed and Hardy-Weinberg expected het- erozygosity). We used single-locus chi-square tests of Hardy-Weinberg equilibrium (HWE) with Le- vene's correction for small sample size and Fisher's 9 10 12 11 13 14 15 16 100 Km FIG. 1. Distribution of Red-legged Partridges (striped areas) and Rock Partridges (dotted areas) in France and Italy, and distribution of sampled populations. The darker area overlapping part of the Rock Partridge distribution in northwestern Italy indicates the distribution of recently extinct Rock Partridge populations in the western Italian Apennines. The reticulated area indicates the distribution of recently extinct Red-legged Partridge populations in Piedmont (Italy). The present hybrid zone is indicated by an oval. The sampled populations are numbered as follows (numbers 1 to 3 are Red-legged Partridges; numbers 5 to 18 are Rock Partridges): 1 = central Portugal; 2 = central Spain; 3 = Cipieres, France (ca. 20 to 24 km from the hybrid zone); 4 = hybrids from the hybrid zone; 5 = Rock Partridges from the hybrid zone; 6 = the high Valley of Tin6e, France (14 to 28 km from the contact zone); 7 = Cuneo, Italy (30 to 40 km from the contact zone); 8 = Champsaur, France (70 to 80 km); 9 = Queyras, France (75 to 85); 10 = Isere, France (80 to 90 km); 11 = Valle d'Aosta, Italy (130 to 150 km); 12 = Savoie, France (170 to 190 km); 13 = Novara, Italy (ca. 250 km); 14 = Sondrio, Italy (ca. 350 km); 15 = Brescia, Italy (ca. 400 km); 16 = Pordenone, Italy (ca. 550 km); 17 = central Apennines, Abruzzo, Italy; 18 = Albania. The pie diagrams indicate the percent of Red-legged Partridge (white) and Rock Partridge alleles (black) found in each population. TABLE 1. Plumage differences between Rock Partridge and Red-legged Partridge. 327 Character Rock Partridge Red-legged Partridge Head color Bar coverts Black eye stripe White line over eye Frontal black band White throat patch Black necklace Nape Flank feathers Axillary feathers Gray Black with buff tips More than 2 mm wide Narrow (1 mm wide) >2 mm wide Maximum height >5 cm Sharply demarcated No streak on nape Two black bars At least one black bar Forehead gray, rest of head brown Yellow-buff Very narrow (<1 mm) Line on forehead broadens into eye stripe Very narrow (<1 mm) or incomplete Smaller (maximum height <5 cm) Spreads into black streaks on upper chest Some black streaks on nape One black bar No black bar exact probability test (Weir 1990). Workman and Nis- wander's (1970) contingency chi-square test was used to assess the significance of allelic heterogene- ity among samples. We computed various genetic- distance matrices (D) and used Rogers' D (1972) to obtain UPGMA (Sneath and Sokal 1973) and neigh- bor-joining dendrograms (NJ; Saitou and Nei 1987). (2) FSTAT and LINKDIS (Black and Krafsur 1985a, b) were used to compute F-statistics (Wright 1965) and to calculate pairwise linkage disequilibrium coeffi- cients within each population estimated as R, the standardized correlation between allele associations. (3) Genetic relationships among populations were represented also through multivariate analyses. NTSYS-pc (Rohlf 1990) was used to obtain the min- imum spanning tree (MST) and non-metric multi- dimensional scaling (MDS) of a principal coordinate analysis (Gower 1966) of the Rogers' D matrix. RESULTS Genetic variability of populations.--Allele fre- quencies (Table 2) diverged significantly be- tween allopatric populations of Red-legged and Rock partridges at Est, Mpi, Got, and Me (Workman and Niswander's contingency test) but not at Amy or Idh. Only one allele, Got-C, was totally absent from one species (Red-leg- ged Partridge) and highly frequent (0.81) in the other (Rock Partridge) and in hybrid pop- ulation 4. Some alleles occurred at low frequen- cy in one species and were found in hybrid population 4 and in populations sampled around the hybrid zone. Est-C, Mpi-B, and Idh- B were absent in the Iberian Red-legged Par- tridge (populations 1 and 2) but frequent in all Rock Partridge populations and present in the hybrid population 4 and in the French Red-leg- ged Partridge (population 3). Got-A, a Red-leg- ged Partridge allele, was present in hybrid pop- ulation 4, in Rock Partridges within the hybrid zone, and in allopatric populations 6, 9, and 10 near the hybrid zone. Amy-C was present at low frequency in Rock Partridge allopatric populations 6 and 8 and in hybrid population 4. Estimates of gene diversity increased toward the hybrid zone, the mean number of alleles per locus (A) was 2.0, P (0.05 criterion) 50%, Ho 0.19 in all the populations sampled around the hybrid zone (3 to 10; see Fig. 1), and the highest values of H o and H,. occurred in hybrid population 4. Heterozygosity was higher in French and Italian Rock Partridges sampled near the hybrid zone than in other allopatric Italian and Iberian populations. These findings suggest introgression of alleles through the hy- brid zone. Significant deviations from HWE (Table 3) were observed for some of the loci in six pop- ulations near the hybrid zone, and in Red-leg- ged Partridges from Portugal. These birds were collected at four localities (Bombarral, Mertola, Campo Major, Castro Marlin) in central and southern Portugal and may represent a hetero- geneous sample with different allele frequen- cies at the different locales (i.e. the Wahlund ef- fect; Hartl and Clark 1990). Nevertheless, they were pooled in a single group because differ- ences in allele frequencies among them have no effects on the description of the genetic clines. Hybrid population 4 was in HWE. All Rock Partridge populations far from the hybrid zone, i.e. east of the Savoie/Valle d'Aosta Alps (pop- ulations 13 to 18; see Fig. 1) were in HWE. Sig- nificant values of the fixation index F were al- ways due to a deficit of heterozygotes from ex- pected (i.e. positive F values; Table 3). The genotypic composition of hybrid and al- lopatric populations was described through the distribution of an individual hybrid index (Bar- ton and Gale 1993): 328 RANDI AND BERNARD-LAURENT [Auk, Vol. 116 �� TABLE 3. Loci showing significant (P < 0.05) devi- ations from Hardy-Weinberg equilibrium and probability values obtained using Levene and Fisher chi-square tests. F is the fixation index (Wright 1965). Locus Levene test Fisher test F Red-legged Partridge-1 (n = 20) Est 0.003 0.030 0.608 Amy 0.000 0.001 0.762 Red-legged Partridge-3 (n --- 63) Me 0.012 0.017 0.312 Rock Partridge-5 (n = 7) Mpi 0.001 0.091 1.000 Got 0.004 ns 0.600 Rock Partridge-6 (n = 72) Me 0.021 ns 0.264 Rock Partridge-7 (n = 53) Idh 0.005 0.017 0.369 Rock Partridge-9 (n = 28) Est 0.020 0.010 0.407 Got 0.000 ns 0.486 Rock Partridge-1 (n = 8) Mpi 0.038 ns 0.590 z = az,, (1) where a = 1/2n, n = the number of diagnostic marker genes, and z, = genotypes generated by the marker genes (indicated as 0, 1, or 2). The individual hybrid index was computed using the frequencies of the most common alleles in the Red-legged Partridge populations (and in the case of loci with three alleles, i.e. the fre- quency of Est-A and not Est-B+C, and so forth); therefore, values of z can range from 0 (for genetically pure Rock Partridges) to 1 (for pure Red-legged Partridges). Allopatric popu- lations of Rock Partridges (Fig. 2A) and Red- legged Partridges (Fig. 2C) had very skewed distributions of z, with modal values of 0 and 0.9, respectively. The distribution of z within hybrid population 4 (Fig. 2B) was not unimodal and therefore did not indicate a predominance of F1 hybrids (the central class with 50% genes derived from both the parents). Nor was it strictly bimodal (indicating the complete ab- sence of F individuals), but there were fre- quent backcrosses (25% and 75% classes) and other recombinants. French Red-legged Par- tridges and Rock Partridges collected near the hybrid zone (not shown) had a high frequency of intermediate genotypes, which suggests al- lelic introgression. These findings are consis- tent with the mixed-phenotype composition of the partridges sampled within the hybrid zone: hybrid, Rock Partridge, and Red-legged Par- tridge phenotypes accounted for 86%, 10%, and 4%, respectively, of the studied partridges (Ber- nard-Laurent 1984). Outside the hybrid zone all birds appeared to be phenotypically pure. Multivariate relationships among populations.- Populations were plotted in a 3-dimensional space (MDS), and a minimum-spanning tree (MST) was superimposed to show relation- ships among the populations (Fig. 3). Allopat- eo A eo C 3o 20 2 20 B ,I.ll ,I Z values (x 100) FIG. 2. Distribution of the frequency (%) of the individual hybrid index based on allozyme frequencies (z x 100) in: (A) eastern Rock Partridges (populations 13 to 18); (B) populations 4 and 5 sampled in the hybrid zone; and (C) western Red-legged Partridges (populations 1 and 2). Intermediate values of z indicate F ge- notypes, whereas extreme values (<25, >75) indicate different backcrosses and recombinant genotypes. rridge populations F!G. 3. Three-dimensional distribution of partridge populations obtained by non-metric multidimension- al scaling of a principal coordinate analysis (Gower 1966) of the Rogers' D matrix. Populations are connected by a minimum spanning tree. See Figure 1 for acronyms and distributions of populations. ric Red-legged and Rock partridge populations were completely separated on opposite sides of the first dimension, and the hybrid population was almost intermediate. The tight cluster of Rock Partridges on the right side of the plot en- compasses all populations east of Valle d'Aosta and the Savoie Alps (populations 13 to 18) and indicates low levels of genetic divergence among populations. On the contrary, the west- ern Rock Partridge populations (numbers 5 to 12) exhibited increased genetic divergence. MDS also depicted the introgression of Red- legged Partridge alleles into the Rock Partridge populations near the hybrid zone (i.e. popula- tions 5 to 7, which were collected near the hy- brid zone, were directly linked to hybrid pop- ulation 4 through the MST; see Figs. 1 and 3). Dendrograms obtained by clustering Rogers' D matrix with UPGMA and NJ procedures were very similar (not shown) and confirmed the re- lationships obtained with MDS. Genetic clines.--Single-locus clines at the four loci that showed significant interspecific diver- gence (Est, Mpi, Got, Me) were represented by plotting frequencies of the most frequent alleles in the Red-legged Partridge populations against distance of each population from the center of the hybrid zone. Genetic clines showed similar shapes and were roughly co- incident, with intermediate and abrupt change of allele frequencies in hybrid population 4. Al- though frequencies of Got-B and Me-B decayed with exponential regularity far from the hybrid zone, the Est-A and Mpi-A alleles showed greater local departures from the curve. In par- ticular, Rock Partridges from Valle d'Aosta and Savoie (populations 11 and 12) exhibited in- creased frequencies of Est-A (Fig. 4A). Other populations deviated from a regular clinal pat- tern, e.g. population 10 had high frequencies of Mpi-A and Got-B (Figs. 4B, 4C), and popula- tion 7 had a high frequency of Me-B (Fig. 4D). Pairwise comparisons of allele frequencies in populations 4 to 18 showed that Est and Mpi alleles introgressed into Rock Partridges sig- nificantly more than Me and Got alleles, and Mpi significantly more than Me and Got alleles (Fisher's exact test at each locality with sequen- tial Bonferroni adjustment; Rice 1989). These allelic fluctuations suggest differential intro- gression of Red-legged Partridge alleles into some Rock Partridge populations, although the observed allelic fluctuations could be due to low sample size (i.e. population 7; n = 8), or to local isolation and random drift (i.e. population 10). We computed the hybrid index: HI = Z(Pl)/ 4 (Vanlerberghe et al. 1988) using the most fre- quent allele (p) in Red-legged Partridges at each of the four discriminating loci (Est-A, 100 -- 80 6O 40- 20-- 41 6 5 - 7 8-1( I -1000 -500 Est-A I 2 8O I 11 I 19i 3 60 40 6-8 15 16 17 18 I 500 lOOO -lOOO -500 (B) Mpi-A 500 1000 6O 40 20- 4 6-9 7-8 I -600 (c) Got-B 10 1112 16 17 18 I 500 1000 100 80 60 40 I I (o) Me-B -1000 -600 0 500 1000 80- (E) Hybrid index I -1000 -500 m 600 1000 FIG. 4. Clinal variation of four diagnostic alleles in Red-legged Partridge, Rock Partridge, and hybrid populations. Allele frequencies (x100) are plotted against distance (km) of each population from the mor- phological hybrid zone (0 km). Populations are plotted from east to west, according to the map in Figure 1. The loci are: (A) Est-A, (B) Mpi-A, (C), Got-B, and (D) Me-B; (E) is the plot of the hybrid index computed following Varlenberghe et al. (1988). Distances between populations are approximate measures from the map. Mpi-A, Got-B, Me-B). Values of HI plotted against distance from hybrid population 4 (Fig. 4E) exhibited a stepped clinal transition from the high frequencies of Red-legged Partridge alleles to the eastern Rock Partridge samples. Hybrid population 4 had HI = 0.53, almost ex- actly intermediate between the HI of Red-leg- ged Partridges far from the hybrid zone (0.98 in Spain and 0.90 in Portugal) and that of east- ern Rock Partridges (0.03 in Albania). The width of the multilocus cline estimated by fit- ting the observed sigmoid plot to a logistic re- gression model was w = 70 to 80 km, whereas it was w = 140 to 160 km when estimated as the distance between HI = 0.78 and 0.22 (corre- sponding to the difference between 0.80 and 0.20, as corrected following May et al. 1975). Cline width indicates that introgression of Red-legged Partridge alleles into Rock Par- tridge populations extends far from the geo- graphic location of the hybrid zone, which is only 15 km wide as assessed through the dis- tribution of hybrid phenotypes. These values of HI in fact included some of the populations sampled west of the Savoie/Valle d'Aosta Alps (i.e. populations 6, 7, and 10; Figs. 1 and 4E). Red-legged Partridge and other Rock Partridge populations in France (i.e. 8, 9, 11, and 12) had HI values very close to 0.78 and 0.22, respec- tively. Therefore, the allozyme cline included some phenotypically pure Rock Partridges as follows: population 6 (collected at ca. 14 to 28 km from the hybrid zone), 7 (collected in Pied- mont, Italy, from the eastern slopes of the Southern Alps, ca. 30 to 40 km from the hybrid zone), 10 (collected in Isere, ca. 80 to 90 km from the hybrid zone), and 12 (collected in Sa- voie, ca. 170 to 190 km from the hybrid zone; see Fig. 1). Introgression of Red-legged Par- tridge alleles into Rock Partridge populations located within 100 km of the center of the hy- brid zone (i.e. populations 6, 7, 8, and 10) in- volved all polymorphic loci, whereas high HI values of the Savoie (12) and Valle d'Aosta (11) populations were due to the high frequency of Est-A alone (Fig. 4A). The allozyme cline was not defined on the Red-legged Partridge side. It was not possible to sample more French Red- legged Partridge populations because the birds are rare toward the hybrid zone, and some- times populations are heavily restocked with captive-raised birds. Linkage disequilibrium.--Analysis of linkage disequilibrium revealed significant R coeffi- cients in five of 18 populations (Table 4) owing to nonrandom positive associations among al- leles derived from the same parental popula- tion. Average pairwise R values (Table 2) were plotted against distances from the hybrid zone (Fig. 5A) and compared with the plot of the ob- served heterozygosities (Fig. 5B). Spacial trends in both estimates corresponded with multilocus and single-locus clines: maximum R and Ho values were in or near the hybrid zone. Two Rock Partridge populations (7 and 10) within the area of introgression had R values higher than the hybrids (Fig. 5A). We cannot TABLE 4. Loci with significant (P < 0.05) values of pairwise linkage disequilibrium. Alleles of each pair are both indicated in capital letters if conspe- cific (e.g. row 1, Got-B and Me-B ar highly frequent alleles in Red-legged Partridges), or one in capital and one in lowercase if heterospecific (row 2, Got- B is a Red-legged Partridge allele, and Me-a is a Rock Partridge allele). Observed = the observed frequency (%) of a given allele association; Ex- pected = expected random frequency (%) of a giv- en allele association; P = probability of occurrence of deviations between observed and random ex- pected allele associations. Loci Alleles Observed Expected P Red-legged Partridge-3 (n = 63) Got / Me B / B 73.5 70.4 0.02 B/a 29.5 32.6 0.02 Hybrids (n = 30) Est/Got Got/Me A/B 17.5 14.7 0.04 A/c 19.0 22.7 0.01 C/b 4.5 7.3 0.04 C/C 15.0 11.0 0.01 B/a 8.5 13.2 0.00 B/B 13.5 8.8 0.00 C/A 24.0 20.4 0.02 C/b 10.0 13.6 0.02 Rock Partridge-5 (n = 7) Est/Me A/A 5.5 4.5 0.02 A/B 2.5 1.5 0.02 C/A 6.5 5.5 0.02 C/b 0.5 1.5 0.02 Rock Partridge-7 (n = 53) Mpi/Got A/B 1.0 0.2 0.00 B/b 0.0 0.8 0.00 Rock Partridge-9 (n = 28) EST/Mpi C/a 10.0 13.5 0.04 C / B 26.0 22.5 0.04 B/A 2.0 0.7 0.01 B/b 0.0 1.2 0.01 rule out an effect of small sample size on esti- mated R at Cuneo (population 7; n = 8), but the Isere sample (population 10; n = 28) was larger than the average sample size of this study (n = 25). DISCUSSION The Alps represent a barrier to dispersal and an area of secondary contact among many southwestern and southeastern taxa (Taberlet et al. 1998). The slopes of the southern French Alps near the Italian border support several hybrid zones, including those formed by two chromosomal races of a grasshopper (Podisma pedestris; Hewitt 1975), two subspecies of crow 35 3O- 25- 2O Km 35O B 4 300 6 7 250 8 150 12 16 10(] i 2 18 Km FIC. 5. Plots of the (A) average coefficient of linkage disequilibrium R (x100), and (B) observed hetero- zygosity (x100) across the partridge hybrid zone. (Corvus corone; Saino et al. 1992), and the two species of partridge described in this paper. The Alps were covered by ice during the last glacial maximum, about 18,000 to 20,000 years ago, and most subalpine areas and the central European plains consisted of tundra (Roberts 1989). Alpine and subalpine habitats were not suitable for partridges, which typically inhabit arid flatlands, open hilly slopes, and rocky de- forested mountains. Following Holocene warm- ing about 10,000 years ago, ice and tundra re- treated, and many plant and animal popula- tions expanded their ranges northward from southern refugia (Hewitt 1989). Biochemical and molecular analyses (Randi et al. 1992, Randi 1996) suggested that Red-leg- ged and Rock partridges speciated in allopatry about 1.5 to 2.0 millions years ago, at the onset of Pliocene/Pleistocene climatic deterioration. Red-legged and Rock partridges probably were confined to meridional refuge areas in Iberia, Italy, France, and the Balkans, at least until 8,000 to 9,000 years ago, when most of Eu- rope and the Alps were covered by forests. Pa- laeoecology of the Alps suggests that deforest- ed areas resulting from climatic changes and human activities were widespread from 6,000 to 5,000 years before present (Hewitt 1993). The spread of prehistoric agriculture may have fa- vored the dispersal of partridges that often use pastures, vineyards, orchards, and other culti- vated lands (Johnsgard 1988). Thus, the present Red-legged Partridge x Rock Partridge hybrid zone probably originated from a secondary contact in the southern French Alps following the Holocene climatic amelioration, not before 8,000 years ago and probably about 6,000 years ago. Steep variation in allele frequencies was cen- tered at hybrid population 4, with a shift of val- ues of the multilocus hybrid index HI by 60% over about 60 km distance. Allozyme clines across the partridge hybrid zone were coinci- dent but only roughly concordant, because al- lele frequencies at some loci fluctuate in popu- lations sampled within about 160 km of the hy- brid zone (Fig. 4). The hybrid population had higher heterozygosity and linkage disequilib- ria than other populations within and near the hybrid zone. Coincident and concordant clines generated by different and non-linked struc- tural loci, as well as linkage disequilibria, are predicted by the tension-zone model of hybrid zones (Barton and Hewitt 1985). The observed allozyme clines in partridges could be main- tained through a balance between selection against hybrid genotypes and migration of pa- rental and recombinant individuals. However, the multimodal distribution of the individual hybrid index z (Fig. 2), due to the presence of different genotypic classes within the hybrid population, suggests that intrinsic selection against F hybrids within the hybrid zone is weak. Moreover, although the hybrid popula- tion was in Hardy-Weinberg equilibrium, other populations near the hybrid zone showed a sig- nificant deficit of heterozygotes and linkage disequilibrium, suggesting that natural selec- tion could eliminate some hybrid genotypes and impede gene flow outside the phenotypic hybrid zone. Variations in the observed shape of the allo- zyme clines could be due to stochastic fluctu- ations (e.g. small size of local populations or small sample size) or to selective barriers against gene flow (Barton and Hewitt 1985, Szymura and Barton 1986, Pialek and Barton 1997). The expected width of a cline, as derived from neutral diffusion of allelic variants, is w = 2.51T 0.5 (Barton and Gale 1993), where is the gene flow computed as the standard deviation of juvenile dispersal, and T is the presumed time in generations since the secondary con- tact. The observed allozyme cline in partridges has a width of w = 70 to 160 km (Fig. 4E). It is difficult to obtain an estimate of , because em- pirical observations are scarce. From Bernard- Laurent's (1991a) observations on movements of 13 radio-tagged juvenile hybrid partridges, short-term dispersal was = 1.0 km, with highly variable individual ranges (0.3 to 25 km). Maximum reported travel distances for Rock Partridges are 25 km in France (Bernard- Laurent, 1991b) and 8 to 9 km in Austria (Haf- ner 1994). Using values of dispersal of 10 to 20 km per generation, and w = 70 to 160 km, we computed values of T = 5 to 40 generations, corresponding to 10 to 120 years if generation time is 2 to 3 years, that are necessary to pro- duce the observed genetic cline, assuming strict neutrality. Accordingly, the observed multilo- cus cline should be very recent, and / or the dis- persal rate very low, if generated by neutral dif- fusion of alldes. In other words, neutral dif- fusion with ranging from 10 to 20 km per generation and T of 2 to 3,000 generations since the secondary contact of the two species should have produced a cline of 1,120 to 2,750 km, which is substantially wider than the observed cline of 70 to 160 km. The theory of tension zones offers more rigorous and formal ap- proaches to the study of genetic clines (Szy- mura and Barton 1986, Mallet et al. 1990, Pialek and Barton 1997), but the incompleteness of our data prevent further elaboration. In particular, we have not sampled more Red-legged Par- tridges in France because they are scarce around the contact zone, and most localities have been heavily restocked with captive- raised birds. Therefore, because only one side of the hybrid zone was properly sampled, the position, shape, and width of the cline cannot be estimated with greater precision. In documenting autumn migratory move- ments of Rock Partridges, Bernard-Laurent (1991b) found that birds were able to fly across the hybrid zone and reach the Red-legged Par- tridge habitat at lower elevations. Migratory behavior and dispersal suggest a model of for- mation and dynamics of the partridge hybrid zone: parapatric populations of partridges are separated by boundaries that are located along an area of equilibrium between a cline of de- creasing adaptability to environmental condi- tions (for Red-legged Partridges at higher ele- vations) and a cline of increasing interspecific competition (for Rock Partridges at lower ele- vations in the habitat of the other species). This boundary can be crossed, for example, by Rock Partridges that penetrate the Red-legged Par- tridge areas where they occasionally stay and breed with the resident Red-legged Partridges. The fertility of hybrids has been directly as- sessed after capturing 13 fertile pairs with hy- brid phenotypes (Bernard-Laurent 1990). The multimodal distribution of z suggests that fer- tile F hybrids can initiate a hybrid population that can backcross and introgress into border- ing populations. The stability of the hybrid zone is sustained by the high reproductive suc- cess of hybrids (Bernard-Laurent 1987), but the diffusion of hybrids outside the narrow hybrid zone toward the Rock Partridge habitat could be limited by higher winter mortality owing to starvation caused by deep snow (Bernard-Lau- rent 1988). Red-legged Partridges and Rock Partridges are widespread, and their distributions are fragmented in patches of allopatric and para- patric populations that are more or less genet- ically isolated through reduced gene flow among them. The homogenizing effects of in- trogression through the hybrid zone seem to be strongly counteracted by natural selection against recombinant genomes, which forces the hybrid zone to remain narrow at the present lo- cation and acts as a strong barrier to gene flow of Red-legged Partridge alleles into Rock Par- tridge populations. ACKNOWLEDGMENTS We thank P. Leonard, D. Dias, I. Artuso, M. Pagan- in, and D. Febbo for help in collecting partridges. P. U. Alkon, A. J. Baker, N. Barton, P. Boursot, E Van- lerberghe, and an anonymous referee provided sug- gestions that greatly improved an earlier version of the manuscript. Laboratory assistance of A. L. Bolelli and V. Lucchini has been deeply appreciated. This study was supported by the Istituto Nazionale per la Fauna Selvatica Ozzano dell'Emilia (BO), Italy, and Office Nationale de la Chasse, France. LITERATURE CITED ARNOLD, M. L., AND S. HODGES. 1995. Are natural hybrids fit or unfit relative to their parents? Trends in Ecology and Evolution 10:67-71. BARROWCLOUGH, G. F. 1980. Genetic and phenotypic differentiation in a wood warbler (genus Den- droica) hybrid zone. Auk 97:655-668. BARTON, N.H., AND K. S. GALE. 1993. Genetic anal- ysis of hybrid zones. Pages 13-45 in Hybrid zones and the evolutionary process (R. G. Har- rison, Ed.). Oxford University Press, New York. BARTON, N. H, AND G. M. HEWITT. 1985. Analysis of hybrid zones. Annual Review of Ecology and Systematics 16:113-148. BARTON, N. H, AND G. M. HEWITT. 1989. Adaptation, speciation, and hybrid zones. Nature 341:497- 503. BERNARD-LAURENT, A. 1984. Hybridation naturelie entre Perdrix bartavelle (Alectoris graeca saxatilis) et Perdrix rouge (Alectoris tufa tufa) dans les A1pes-Maritimes. Gibier Faune Sauvage 2:79-96. BERNARD-LAURENT, A. 1987. D6mographie compa- r6e d'une population de Perdrix bartavelle (A. g. saxatilis) et d'une population d'hybrides (A. g. saxatilis x A. tufa tufa) dans les A1pes-Mari- times. Revue d'Ecologie (Terre Vie) Supplement 4:189-197. BERNARD-LAURENT, A. 1988. Les dplacement en au- tomne et en hiver de Perdrix rochassire (Alec- toris graeca saxatilis x Alectoris tufa tufa) dans les Alpes mridionales et leurs dterminants. Gi- bier Faune Sauvage 5:171-186. BERNARD-LAURENT, A. 1990. Biologie de reproduc- tion d'une population de Perdrix rochassire Alectoris graeca saxatilis x Alectoris tufa tufa dans les Alpes m6ridionales. Revue d'Ecologie (Terre Vie) 45:321-344. BERNARD-LAURENT, A. 1991a. Structure sociale et utilisation de l'espace par la Perdrix rochassire: Variations saisonnires et individuelles. Gibier Faune Sauvage 8:1-30. BERNARD-LAURENT, A. 1991b. Migrant Rock Par- tridges in the southern French Alps. Journal ffir Ornithologie 132:220-223. BLACK, W. C., AND E. S. KRAFSUR. 1985a. A Fortran program for analysis of genotypic frequencies and description of the breeding structure of populations. Theoretical and Applied Genetics 70:484-490. BLACK, W. C., AND E. S. KRAFSUR. 1985b. A Fortran program for the calculation and analysis of two- locus linkage disequilibrium coefficients. Theo- retical and Applied Genetics 70:491-496. BROWN, M. J., AND M. B. ROBBINS. 1986. Extensive protein similarity of the hybridizing chickadees Parus atricapillus and P. carolinensis. Auk 103:667- 675. COOK, A. 1975. Changes in the Carrion/Hooded crow hybrid zone and the possible importance of climate. Bird Study 22:165-168. CORBIN, K. W., AND P. J. WILKIE. 1988. Genetic simi- larities between subspecies of the White- crowned Sparrow. Condor 90:637-647. DECLAND, D., AND Z. GERBE. 1867. Ornithologie Eu- ropeenne. Baillre, Paris. EMMS, K. S., AND M. L. ARNOLD. 1997. The effect of habitat on parental and hybrid fitness: Trans- plant experiments with Louisiana irises. Evolu- tion 51:1112-1119. ENDLER, J. A. 1977. Geographic variation, speciation, and clines. Princeton University Press, Prince- ton, New Jersey. GILL, EB. 1994. Rapid, asymmetrical introgression of mitochondrial DNA in hybridizing popula- tions of Blue-winged and Golden-winged war- biers. Journal ffir Ornithologie 135:359. GILL, EB. 1997. Local cytonuclear extinction of the Golden-winged Warbler. Evolution 51:519-525 COWER, J. C. 1966. Some distance properties of latent root and vector methods used in multivariate analysis. Biometrika 53:325-338. GRANT, P. R., AND B. R. GRANT. 1992. Hybridization of bird species. Science 256:193-197. GRUDZIEN, T. A., AND W. S. MOORE. 1987. Genetic dif- ferentiation between the Yellow-shafted and the Red-shafted subspecies of the Northern Flicker. Biochemical Systematics and Ecology 14:451- 453. HAFNER, F. 1994. Das Steinhuhn in Karnten. Okolo- gie, Verhalten und Lebenstraum. Sonderheft 52 der "Carinthia II", Naturwissenachaftlicher Ver- ein ffir Karnten, Klagenfurt, Austria. HARE, M. I, AND J. C. AVISE. 1996. Molecular genetic analysis of a stepped multilocus cline in the American oyster (Crassostrea virginica). Evolu- tion 50:2305-2315. HARRISON, R. G. 1986. Pattern and process in a nar- row hybrid zone. Heredity 56:337-349. HARRISON, R. G. 1990. Hybrid zones: Windows on evolutionary process. Oxford Surveys in Evolu- tionary Biology 7:69-128. HARRISON, R. G. 1993. Hybrids and hybrid zones: Historical perspective. Pages 3-12 in Hybrid zones and the evolutionary process (R. G. Har- rison, Ed.). Oxford University Press, New York. HARTL, D. L., AND A. G. CLARK. 1990. Principles of population genetics. Sinauer Associates, Sun- derland, Massachusetts. HEWITT, G. M. 1975. A sex chromosome hybrid zone in the grasshopper Podisma pedestris (Orthop- tera: Acrididae). Heredity 35:375-385. HEWITT, G. M. 1989. The subdivision of species by hybrid zones. Pages 85-110 in Speciation and its consequences (D. Otte and J. A. Endler, Eds.). Sinauer Associates, Sunderland, Massachusetts. HEWITT, G. M. 1993. After the ice: Parallelus meets er- ythropus in the Pyrenees. Pages 140-164 in Hy- brid zones and the evolutionary process (R. G. Harrison, Ed.). Oxford University Press, New York. HOWARD, D. J., AND G. L. WARINC. 1991. Topograph- ic diversity, zone width, and the strength of re- productive isolation in a zone of overlap and hy- bridization. Evolution 45:1120-1135. JOHNSCARD, P. A. 1988. The quails, partridges, and francolins of the world. Oxord University Press, New York. KEY, K. H. L. 1968. The concept of stasipatric speci- ation. Systematic Zoology 17:14-22. MALLET, J., N. BARTON, G. M. LAMAS, J. C. SANTIS- WEBAN, M. M. MUEDAS, AND H. EELEY. 1990. Es- timates of selection and gene flow from mea- sures of cline width and linkage disequilibrium in HeIiconius hybrid zones. Genetics 124:921- 936. MAY, R. M., J. A. ENDLER, AND g. E. MCMUTRIE. 1975. Gene frequency clines in the presence of selec- tion opposed by gene flow. American Naturalist 109:659-676. MOORE, W. S. 1987. Random mating in the Northern Flicker hybrid zone: Implications for the evolu- tion of bright and contrasting plumage patterns in birds. Evolution 41:539-546. MOORE, W. S., AND D. BUCHANAN. 1985. Stability of the Northern Flicker hybrid zone in historical times: Implications for adaptive speciation the- ory. Evolution 39:135-151. MOORE, W. S., J. H. GRAHAM, AND J. T. PRICE. 1991. Mitochondrial DNA variation in the Northern Flicker (Colaptes auratus, Aves). Molecular Biol- ogy and Evolution 8:327-344. PIALEK, J., AND N.H. BARTON. 1997. The spread of an advantageous allele across a barrier: The effects of random drift and selection against heterozy- gotes. Genetics 145:493-504. PR1GOG1NE, A. 1980. Etude de quelque contacts se- condaires au Zaire oriental. Le Gerfault 70:305- 384. RAND, D. M., AND R. G. HARRISON. 1989. Ecological genetics of a mosaic hybrid zone: Mitochondrial, nuclear and reproductive differentiation of crickets by soil type. Evolution 43:432-439. RANDI, E. 1996. A mitochondrial cytochrome b phy- logeny of the Alectoris partridges. Molecular Phylogenetics and Evolution 6:214-227. RANDI, E., A. MER1GG1, R. LORENZINI, G. Fus1/2o, AND P. U. ALKON. 1992. Biochemical analysis of re- lationships of Mediterranean Alectoris partridg- es. Auk 109:358-367. RICE, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225. RISING, J. D. 1983. The Great Plains hybrid zones. Current Ornithology 1:137-157. ROBERTS, N. 1989. The Holocene: An environmental history. Basil Blackwell, Oxford. ROGERS, J. S. 1972. Measures of genetic similarity and genetic distances. Studies in Genetics (Univer- sity of Texas Publications) 7213:145-153. ROHLF, J. E 1990. NTSYS-pc. Numerical taxonomy and multivariate analysis system, version 1.7. Exeter Software, Setauket, New York. SAINO, N., R. LORENZINI, G. Fusco, AND E. RANDI. 1992. Genetic variability in a hybrid zone be- tween Carrion and Hooded crows (Corvus cotone cotone and C. c. cornix, Passeriformes, Aves) in north-western Italy. Biochemical Systematics and Ecology 20:605-613. SAITOU, N., AND M. NEI. 1987. The neighbor-joining method: A new method for reconstructing phy- logenetic trees. Molecular Biology and Evolution 4:406-425. SEARLE, J. B. 1993. Chromosomal hybrid zones in eu- therian mammals. Pages 303-353 in Hybrid zones and the evolutionary process (R. G. Har- rison, Ed.). Oxford University Press, New York. SHORT, L. H. 1969. Taxonomic aspects of avian hy- bridization. Auk 86:84-105. SITES, J. W., JR., N.H. BARTON, AND K. M. REED. 1995. The genetic structure of a hybrid zone between two chromosornal races of the Sceloporus gram- micus complex (Sauria, Phrynosomatidae) in central Mexico. Evolution 49:9-36. SNEATH, P. H., AND R. R. SOKAL. 1973. Numerical tax- onomy. W. H. Freeman, San Francisco. SPAN), S. 1979. Nuovi ibridi naturali Alectoris rufa tufa (L.) x Alectoris graeca saxatilis (Bechstein) sulle Alpi Marittime e relative considerazioni tassonomiche. Annali del Museo Civico di Storia Naturale di Genova 82:154-162. SWOFFORD, D. L., AND g. K. SELANDER. 1989. BIOSYS- 1. A computer program for the analysis of allelic variation in population genetics and biochemical systematics. Release 1.7. Illinois Natural History Survey, Champaign. SZYMURA, J. M., AND N.H. BARTON. 1986. The genetic structure of the hybrid zone between the fire- bellied toads Bombina bombina and B. variegata, near Cracow in southern Poland. Evolution 40: 1141-1159. TABERLET, P., L. FUMAGALLI, A.-G.WuST-SAUCY, AND J.-E COSSON. 1998. Comparative phylogeogra- phy and postglacial colonization routes in Eu- rope. Molecular Ecology 7:453-464. VANLERBERGHE, F., P. BOURSOT, J. CATALAN, S. GER- AS1MOV, F. BONHOMME, B. A. BOTEV, AND L. THALER. 1988. Analyse gntique de la zone d'hybridation entre les deux sous-espces de souris Mus musculus domesticus et Mus musculus muscuIus en Bulgarie. Genome 30:427-437. WEIR, B. S. 1990. Genetic data analysis. Sinauer As- sociates, Sunderland, Massachusetts. WORKMAN, 13. L., AND J. D. NISWANDER. 1970. Popu- lation studies on southwestern Indian tribes. II: Local genetic differentiation in the Papayo. American Journal of Human Genetics 22:24-29. WRIGHT, S. 1965. The interpretation of population structure by F-statistics with special regards to systems of mating. Evolution 19:395-420. Associate Editor: A. J. Baker