Molecular Ecology (2007) 16, 1207–1219

doi: 10.1111/j.1365-294X.2007.03231.x

Fine-scale spatial genetic structure in mixed oak stands with different levels of hybridization

Blackwell Publishing Ltd

M . V A L B U E N A - C A R A B A Ñ A ,* S . C . G O N Z Á L E Z - M A R T Í N E Z ,† O . J . H A R D Y ‡ and L . G I L * *Unidad de Anatomía, Fisiología y Genética, ETSIM, Ciudad Universitaria s/n, 28040 Madrid, Spain, †Departamento de Sistemas y Recursos Forestales, Centro de Investigación Forestal, CIFOR-INIA, Carretera de La Coruña km 7.5, 28040 Madrid, Spain, ‡Laboratoire Eco-Ethologie évolutive, Faculté des Sciences, Université Libre de Bruxelles, Avenue F.D. Roosevelt 50, Campus du Solbosch, 1050 Brussels, Belgium

Abstract Oaks are model species for the study of natural introgressive hybridization. High interfertility among oak taxa might result in collective evolution, through transpecific spread of advantageous alleles, challenging the standard concept of species. Nine highly polymorphic microsatellite (nuSSR) loci were analysed in three mixed oak populations of Quercus pyrenaica and Quercus petraea (Montejo, Somosierra and Robregordo) with different density and hybridization rates. Both leaf morphology and molecular markers were used to assess individual admixture rates. Insights about the relative effect of density and hybridization rates on fine-scale spatial genetic structure (SGS) were obtained from autocorrelograms and Sp statistics. Differences in SGS among populations were higher than between species. These differences cannot be attributed solely to census densities but also relate to other factors, such as the spatial configuration of the population. Hybridization was an important factor shaping within-population spatial genetic structure, and an interspecific component of SGS was found in Somosierra. Indirect estimates of historical gene flow in Montejo were compared with actual values of gene dispersal assessed by parentage analysis in a former study. Similar values were found for current and historical gene flow in both species, which might reflect demographical stability. Keywords: fine-scale spatial genetic structure, historical gene dispersal, hybrid zones, introgression, microsatellites, Quercus Received 22 May 2006; revision accepted 13 November 2006

Introduction The formation of spatial genetic structure is a crucial feature in plant evolutionary processes and population dynamics (Epperson 2003; Rousset 2004). Given isolation or a reduced effective population size, spatial segregation can lead to speciation through adaptation to different environments (Endler 1977). Spatial aggregation of related genotypes can also promote biparental inbreeding, leading to a reduced fitness due to inbreeding depression (Levin 1984; Campbell 1986; Latta et al. 1998; Stacy 2001). Moreover, the nonrandom distribution of genotypes in space determines an allele-frequency change from one generaCorrespondence: Luis Gil, Fax: (34) 91 543 95 57; E-mail: [email protected] © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

tion to the next, which might affect patterns of natural selection in the long term. In turn, demography and mating systems, together with selection, shape the finescale spatial structure of genetic variation found within populations (Rousset 2004). Mating systems have a major influence on the formation of fine-scale genetic structure. Limited dispersal of propagules leads to progeny clustering around mother trees (e.g. Linhart et al. 1981; Dow & Ashley 1996; GonzálezMartínez et al. 2002). Assortative mating (i.e. the preferential mating of individuals with similar phenotypes) and pollen limitation also favour the clustering of like individuals. After propagule dispersion, microenvironmental selection, if present, can alter fine-scale spatial genetic structure by purging unrelated nonadapted genotypes or favouring kinindividuals from well-adapted families (Epperson 1990).

1208 M . V A L B U E N A - C A R A B A Ñ A E T A L . Assuming restricted dispersal and the isolation by distance model, population structure can provide indirect estimates of historical migration rates (Wright 1943; Malécot 1950; Epperson 1993; Rousset 1997, 2001; Whitlock & McCauley 1999). Furthermore, the observed spatial genetic structure within populations can be used to estimate dispersal parameters (Rousset 1997, 2000; Hardy & Vekemans 1999; Epperson 2005). Theoretical models of isolation by distance in a two-dimensional space predict relatedness to decrease linearly with the logarithm of the distance at a rate that is inversely proportional to 4πDeσ2 (or Nb, the Wright’s formulation of neighbourhood size, conceptually reinterpreted by Rousset 1997), where De is the effective population density and σ2 the variance of axial dispersion (i.e. the average squared dispersal distance measured on an axial scale) from adult to offspring. Effective density is affected by the variance in reproductive success and the individuals’ distribution (Doligez et al. 1998; Meagher & Vassiliadis 2003; but see Epperson 2005). Estimates of the variance of axial dispersion can be obtained either by direct (e.g. parentage analyses; see review in Jones & Ardren 2003) or indirect (e.g. the neighbourhood model, Burczyk et al. 2002; and two-generation analysis or twogener, Smouse et al. 2001; Sork et al. 2002) methods. These methods provide contemporary estimates of dispersal parameters, whereas approaches based on spatial genetic structure represent historical gene movement. A comparison between historical and contemporary gene dispersal can shed light on population dynamics and the change of dispersal parameters at the population level over time (e.g. Dutech et al. 2005). In natural plant populations, hybridization is a common feature. Not only the presence of hybrids but also the complexity of the hybrid system, which might include backcrosses and complex hybrid forms, depend on genetic, physiological and environmental characteristics (see reviews in Carney et al. 2000; Howard et al. 2003). Hybrid occurrence is also affected by the phylogenetic distance among species, which is probably due to higher mating system compatibility among closely related species (Belahbib et al. 2001; Boavida et al. 2001). When hybridizing species co-occur, the establishment of hybrid swarms can result in preferential mating among pure-bred or introgressed types, causing great variance in reproductive success among individuals (according to similar phenotypes, phenology or genetic background). Differences in mating success within and among hybrid classes can influence spatial genetic structure in two opposing ways. On the one hand, an increased variance in reproductive success might decrease effective densities leading to a stronger genetic spatial structure; on the other, given small effective numbers of one hybridizing species, mating with another would increase effective density leading, in this case, to weaker spatial genetic structure.

The amount and spatial pattern of hybrid recruitment can also affect fine-scale spatial genetic structure. In the long term, this effect would depend on hybrid fitness. If hybrid establishment is hindered by ecological factors, reproductive success would be higher in pure populations where all mates are conspecific. Conversely, if the hybrids have equal or higher fitness than the parental species, sympatric hybridizing species may increase their reproductive success due to higher survival rates (e.g. the evolutionary novelty model; Arnold 1997) and the ability to colonize new ranges (e.g. the bounded superiority model; Moore 1977). In the latter case, given some genetic differentiation among the hybridizing species, the general level of withinpopulation relatedness would be lower than in pure stands due to the interspecific crosses. Moreover, the scale and spatial distribution of intra and interspecific mating would considerably affect fine-scale genetic structure patterns.

Aims of the study The local genetic structure of two hybridizing oak species (Quercus pyrenaica and Q. petraea) that differ both in population size and ecological requirements was studied. In central Spain, melojo oak (Q. pyrenaica Willd.) is abundant and continuously distributed, whereas sessile oak [Q. petraea (Matt.) Liebl.] is at its southernmost European range, forming low-density scattered populations. Three mixed populations differing in density and hybrid occurrence were analysed in relation to local spatial genetic structure and historical dispersal rates. More specifically, this study addresses the following questions: (i) are there differences in fine-scale spatial genetic structure between Q. petraea and Q. pyrenaica? A parentage analysis previously performed in one of the populations studied (Montejo; Valbuena-Carabaña et al. 2005) showed contrasting dispersal patterns for each species. Specific dispersal capabilities and differences in population size might be reflected in a stronger local genetic structure in Q. petraea; (ii) are hybrids spatially structured, and how does the amount and distribution of hybrids influence spatial genetic structure of parental species? To answer these questions, first, both morphological traits and molecular markers were used to characterize the amount and distribution of hybrids in each population and second, spatial autocorrelograms were constructed within and between species, with or without the inclusion of putative hybrids at different probability levels; and (iii) are contemporary estimates of gene dispersal in Montejo (see Valbuena-Carabaña et al. 2005 for details) similar to historical estimates based on population structure? A comparison of contemporary and historical dispersal parameters was made to evaluate changes in dispersal processes during the last generation, when Montejo mixed © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

S G S I N O A K S T A N D S W I T H D I F F E R E N T L E V E L S O F H Y B R I D I Z A T I O N 1209 oak forest changed from a dehesa intensively exploited for cattle-grazing, to a unesco Biosphere Reserve.

Materials and methods Study area and populations description The study populations are located on the southern slopes of the Somosierra Massif (Central System Range, northern Madrid, Spain) (Fig. 1a). Three oak mixed-stands with different hybridization rates (as indicated at this initial stage by visual inspection of buds and leaves), from very low (Montejo) to high (Somosierra), were selected. The Central System Range interrupts the arid Castilian Plateau and serves as a contact area between Atlantic and Mediterranean floral elements. Not only climatic but also historical factors related to the intensive use of the land by humans are responsible for species distribution and landscape physiognomy in the region. Nowadays, agrosilvopastoral systems called dehesas (low-density forests managed to produce pasture and forage for livestock) still abound in the mountains of northern Madrid. Ownership of dehesas is communal, corresponding to nearby villages. Typical oak dehesas combine open woodlands with meadows and coppice forests that were used in the past to

produce charcoal. In many cases, the multifunctional exploitation of these lands helped to preserve species and landscapes of high natural and cultural value. It is indeed within these dehesas where the largest and best-preserved temperate oak woodlands are found nowadays, typically forming scattered and isolated populations, either mono or multispecific, and often included in nature reserves. In the Madrid region, Q. petraea (sessile oak), along with companion tree species such as Fagus sylvatica (only in Montejo), Ilex aquifolium, Prunus avium, Sorbus aria, Sorbus aucuparia and Taxus baccata, represents a relict temperate forest. In contrast, Q. pyrenaica (melojo oak) is associated with sub-Mediterranean mountainous ecosystems, broadly distributed in the southern slopes of the Central System Range, where dry summers and late-spring ice favour their presence. Quercus pyrenaica is well adapted to Mediterranean ecological requirements. In addition, its root-sprouting capability is a great advantage against disturbances such as fire or cuttings, with coppice forests — regularly exploited for wood production in the region — being common in this species. The study plots are located in open oak woodlands, where old trees (200–300 years old) are surrounded by clumps of saplings of different age and density, as a result of a notable microenvironmental variability. Oaks are the

Fig. 1 Location map: (a) of the three study populations (Montejo, Somosierra and Robregordo) in the Somosierra Massif (Central System Range, northern Madrid); (b) notice the scattered distribution of trees in the Robregordo population; (c) topographical profiles across the Lozoya Valley showing the different exposure of Robregordo-Somosierra (black line) and Montejo (grey line).

© 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

1210 M . V A L B U E N A - C A R A B A Ñ A E T A L . dominant species in all studied dehesas, though there are differences in relative density of sessile and melojo oaks and in abundance of other temperate trees. Furthermore, contact areas between oak species are more frequent in some populations than in others, probably due to spatial segregation in response to microenvironmental variation (as reported by Pardo et al. 2004 in Montejo) and the different ecological requirements of both species. In Montejo (13 ha sampled; Q. petraea-like N = 92; Q. pyrenaica-like N = 84), the temperate sessile oak grows at a higher altitude and in deeper soils than melojo oak, with contact areas between species being relatively scarce. No morphologically intermediate individuals were found here and the relative density of adults (summer 2005) was 7.02 individuals (ind)/ha and 6.26 ind/ha for Q. petraea and Q. pyrenaica, respectively. In Somosierra (20 ha sampled; Q. petraea-like N = 52; Q. pyrenaica-like N = 118), sessile and melojo oaks intermingle and multiple hybrid forms are present. Density of adults reached higher values than in Montejo, namely 46.95 ind/ha and 52.52 ind/ha for Q. petraea and Q. pyrenaica, respectively. In Robregordo (15 ha sampled; Q. petraea-like N = 84; Q. pyrenaica-like N = 3), Q. petraea dominates over Q. pyrenaica with densities of adult trees of 37.40 ind/ha for Q. petraea and 5.57 ind/ha for Q. pyrenaica. In this location, Q. petraea often grows within holly tree (Ilex aquifolium) and common whitebeam (Sorbus aria) assemblies, which provide intense shade and protection from cattle and game grazing. In contrast, the scarce Q. pyrenaica grows in open and scrub land. Hybrids with intermediate morphology are also found at this location, although at a lower rate than in Somosierra.

Morphological measurements Q. petraea and Q. pyrenaica belong to the same section (Lepidobalanus) but they are reported in the literature as being easily distinguishable based on the morphological traits of leaves, buds and acorns. While Q. petraea presents round-shaped leaves, from glabrous to slightly hairy, Q. pyrenaica has lobulated leaves, with deep sinuses and big lobes densely covered by hair, which is also more abundant and persistent in the lower surface of the blade. Individuals with intermediate morphology have been described in broad contact areas, and the hybrid type was named Q. × trabutii (Hy 1895; Q. × legionensis Vicioso 1950). In the study sites, we have found contrasting numbers of individuals with intermediate morphology: in Somosierra and Robregordo multiple and very variable intermediate morphologies were present, whereas these were absent in Montejo. Interestingly, despite the lack of intermediate morphologies in Montejo, previous studies using molecular markers found ∼8.5% of genetically introgressed individuals in this population (Valbuena-Carabaña et al. 2005).

To develop a discrimination rule between pure individuals and hybrids, 15 morphological traits were measured in five leaves per tree, each one collected from a different branch from the sun-exposed upper part of the crown, in 170 oak trees from Somosierra and 115 from Robregordo (mid-summer 2005). A representative sample of 29 trees was also collected in Montejo to serve as a pure-species reference. Measurements of area (A), perimeter (P), length (L), form coefficient (4πA/P2), width (W), maximal width (MW), position of maximal width (PMW), width at 50% of leaf length (W1), width at 90% of leaf length (W2), maximum sinus depth related to length (MSD/L) and maximum lobe length (MLL) were made with an image analyser software, winfolia version 2002a (Regent Instrument Inc., Canada), using dry leaves. Ratios of P/A and (LML/MSD)/L were also calculated. In addition, hirsuteness (Kissling’s gradient) of the upper (PU) and lower (PL) surface of an intervein blade piece of 2 mm2 was scored using a 10 × microscope.

DNA extraction and molecular markers Leaves from each of 456 oak trees (Montejo N = 176, Somosierra N = 180, Robregordo N = 100) were collected and stored at −80 °C prior to DNA extraction. About 0.5 g of leaf material was ground to a fine frozen powder in liquid nitrogen, from which DNA was extracted. We followed a slightly modified protocol from Doyle & Doyle (1990), using a different detergent (β-ME instead of CTAB) and adding an extra cleaning step with dehydrated EtOH. Nine nuclear microsatellites were scored. Three microsatellite loci (QpZAG9, QpZAG36 and QpZAG110) were developed by Steinkellner et al. (1997) for Q. petraea, four other loci (QrZAG5, QrZAG7, QrZAG11 and QrZAG39) by Kampfer et al. (1998) for Q. robur and two more (MSQ4 and MSQ13) by Dow et al. (1995) for Q. macrocarpa. The PCR was carried out in a thermal cycler (GeneAmp® PCR System 9700, Applied Biosystems, California), using 0.4 units of Ecogen Taq DNA polymerase and approximately 5 ng of genomic DNA in a total volume of 10 µL. The PCR mix also contained 0.2 µm of each primer (forward primers were labelled with the infrared fluorescent IRDye 800 on the 5′ end, purchased from MWG Biotech), 0.2 mm dNTPs, 10 mm Tris/HCl pH 9 and 2 mm MgCl2. After a preliminary denaturation at 94 °C for 5 min, PCR reactions were performed for either 25 cycles (1 min at 94 °C, 30 s at the annealing temperature of 50 °C and 1 min at 72 °C) for QpZAG9, QpZAG110, MSQ4 and MSQ13, or for 30 cycles with an annealing temperature of 53 °C for QrZAG5, QrZAG7, QrZAG11 and QrZAG39. The amplification conditions for QpZAG36 were slightly different: MgCl2 concentration was 3 mm and PCRs were performed for 45 s at 94 °C, 45 s at the annealing temperature of 48 °C and 45 s at 72 °C. Electrophoresis and scoring of fragments © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

S G S I N O A K S T A N D S W I T H D I F F E R E N T L E V E L S O F H Y B R I D I Z A T I O N 1211 were performed on a 4200 LI-COR automated DNA sequencer (LI-COR Biosciences, Lincoln, NE, USA), using a 1 × TBE running buffer, with run parameters of 1500 V, 40 mA and 45 °C plate temperature.

criterion of discrimination between species was established using half the Mahalanobis distance between species and Bayes’ Theorem to construct a posterior probability:

Statistical analyses

Pr(j|x) =

Identification of hybrids Leaf-morphology traits were used to build a discrimination rule between Q. petraea and Q. pyrenaica and their hybrids, using multivariate techniques based on individual tree means (five leaves per tree and 285 trees). Trees collected in Montejo, although useful as a pure-species reference in previous stages of the analyses, were not included in the construction of the final discrimination rule because Q. petraea showed a marked phenotypical differentiation from the other two populations (Somosierra and Robregordo), located only a few kilometres from each other. Variables were transformed when necessary, to improve their statistical properties. Principal component analysis (PCA) was used to summarize all traits into four orthogonal factors that explained ∼90% of the total data variation (see Fig. 2). A Varimax rotation was applied. Using the selected principal components, a canonical discriminant analysis was performed using the DISCRIM procedure implemented in sas version 9.0 (SAS Institute Inc., NC, USA). The discriminant linear function was constructed using the Mahalanobis distance between species means in the covariance matrix of data. An a priori

Fig. 2 Principal component analysis (PCA) of Quercus petraea and Q. pyrenaica in Somosierra-Robregordo based on leafmorphology. PC1 (∼33% of explained variance) is related to lobulation and PC2 (∼5% of explained variance) to pubescence. Circles and triangles represent Q. petraea and Q. pyrenaica individuals, respectively. © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

exp[−0.5D2j (x)]

∑ exp[−0.5Dk2 (x)]

(1)

k

where Pr(j|x) is the posterior probability of individual j to belong to species x, D2j is the generalized squared Mahalobis distance of individual j to the species mean, and Dk2 represents the overall distance of k individuals to the species mean. This formula was used to calculate canonical variate scores for each individual and to reclassify them into pure types or putative hybrids showing intermediate leaf-morphology. Assignation of trees to pure types or hybrids was performed considering two probability levels: a standard 0.95 probability of being a pure type and a more relaxed 0.85 probability that would also consider as pure types slightly introgressed morphologies. In addition to morphological traits, molecular marker scores (nine nuSSRs) were also used to distinguish between pure types and hybrids. A Bayesian model-based clustering method (structure version 2 software, Pritchard et al. 2000), as described in detail in Valbuena-Carabaña et al. (2005) was used to assign individuals (probabilistically) to species. This method allows for the presence of admixed individuals in the species clusters, giving them a posterior probability of belonging to one or the other species. Besides this approximation, hybrids were also identified using the newhybrids program (Anderson & Thompson 2002), obtaining very similar results (data not shown). Following Vähä & Primmer (2005), only structure results are presented here. Data was analysed separately for Montejo (as in Valbuena-Carabaña et al. 2005; but with nine nuSSRs instead of five) and Somosierra-Robregordo populations. The highest posterior probabilities were obtained for K = 2 in both cases, which agrees with the presence of two different species. As in the case of morphological traits, individuals were assigned to pure types or hybrids at two probability levels, the standard 0.95 and a relaxed one of 0.85 (see above). Morphological and molecular-marker individual assignations to pure types or hybrids were combined to classify trees into groups with different levels of introgression. The standard criterion accounted for individuals classified as pure at 0.95 probability using both morphological and genetic methods. The relaxed criterion accounted for individuals with a weaker assignation to their species (0.85 probability of being a pure type), whatever the method used, thus including both pure and slightly introgressed trees. Finally, whole-species groups that included all individuals of a given species (except those with incongruent

1212 M . V A L B U E N A - C A R A B A Ñ A E T A L . morphological and molecular-maker discrimination), irrespectively of their introgression level, were constructed.

Results Identification of hybrids

Fine-scale spatial genetic structure (SGS) Fine-scale spatial genetic structure refers to the decrease of pairwise relatedness with distance within continuous populations (i.e. the isolation-by-distance model, Wright 1943; Malécot 1950). Genetic coancestry between pairs of individuals was computed using J. Nason’s kinship estimator (see Loiselle et al. 1995 for formulae). Average pairwise kinship coefficients were computed for 25 m distance classes, except for Q. petraea in Somosierra, where the five shorter distance classes were at intervals of 15 m. For each species and population, three autocorrelograms were constructed, one considering only pure-bred individuals (at 0.95 posterior probability) and two others including individuals with different introgression levels (at 0.85 posterior probability and considering all sampled trees). SGS significance was tested by permuting 10 000 times spatial locations of individuals. For melojo oak, allele frequencies were estimated removing replicated genotypes, as they probably had clonal origin (see Valbuena-Carabaña et al. 2005). In addition, in Montejo and Somosierra populations, autocorrelograms for interspecific pairs were also computed, using reference gene frequencies from pooled data. Under Wright’s isolation-by-distance model, pairwise kinship comparisons between individuals are expected to vary linearly with the logarithm of distance in a twodimensional space (Rousset 1997, 2000; Hardy & Vekemans 1999; Hardy 2003), at least within some distance ranges (between σ-20 σ, considering a mutation rate for microsatellite markers of µ = 10−3 as in Heuertz et al. 2003). At this spatial scale, the slope of the regression of coancestry coefficients (Fij) with the natural logarithm of distance, blog, equals (FN – 1)/(4πDeσ2) (Vekemans & Hardy 2004), where FN is the kinship between competing individuals; FN can be approximated by F1, which is the coancestry between neighbouring individuals. In all spatial analyses b-log was computed and its significance tested using a randomization procedure whereby individuals of each category were permuted among locations 10 000 times. Vekemans & Hardy (2004) developed a statistic, Sp = − (blog)/1−F1, that avoids biases due to sampling effects. In this study, Sp statistic was used to compare SGS of Q. petraea and Q. pyrenaica growing at different densities and hybridization rates. Sp confidence intervals were obtained from b-log standard errors by jackknifing over loci (Dutech et al. 2005). Finally, an iterative procedure was used in Montejo to obtain σ from b-log estimates assuming 0.9, 0.5 and 0.25 De/D ratios (Ishihama et al. 2005). All the spatial genetic structure analyses were performed using spagedi version 1.2 software (Hardy & Vekemans 2002).

In general, both leaf-morphology (see PCAs in Fig. 2) and molecular markers provided similar estimates of hybridization rates at the population level, although genetic analyses estimated a higher admixture for individual trees. In Montejo, 11 (∼6%) putative hybrids (probability of being pure types lower than 0.85) and 13 (∼7%) slightly introgressed oaks (probability of being pure types between 0.85 and 0.95) were identified using nine nuclear microsatellites. The ability of the method to detect hybridization in Montejo was tested by simulation. One hundred simulated populations (N = 1000 for each species) were constructed using allele frequencies based on pure-bred Montejo’s individuals. structure software was then used to detect hybrids in these simulated datasets. On average, only 1.61% (SD = 0.26) of trees showed any admixture in the simulated populations, pointing out that the level of hybridization found in Montejo was not the result of combination by chance of allele variants that exist in both species (but usually at different allele frequencies). Trees with intermediate leaf-morphology were not found in Montejo. It is noteworthy that increasing the number of nuSSRs from five (in Valbuena-Carabaña et al. 2005) to nine (this study) resulted in ∼4% lower hybridization rate estimates. In contrast to Montejo, Somosierra showed hybridization rates based on molecular markers of ∼22% and also higher proportions of slightly introgressed individuals (∼26%). A substantial number of trees with intermediate leaf-morphology were also found in this population, being ∼18% of trees classified as putative hybrids and ∼14% as slightly introgressed individuals. Overall, the number of putative hybrids in Somosierra was about four-fold that in Montejo. In Robregordo, the hybridization rate was slightly higher than in Montejo (∼9% and ∼6% considering molecular makers and morphology, respectively). Nevertheless, the amount of sessile oaks classified as slightly introgressed was substantially higher in Robregordo (∼2% based on morphological traits but ∼25% based on molecular markers). At the individual-tree level, moderate differences between classification methods in the estimation of introgression rates were found. Furthermore, some individuals were incongruently assigned to different species by the two methods, in particular in Somosierra. In this population, ∼57% of individuals differed in their introgression level (considering the three categories used for SGS analyses) according to morphology and genetic markers, and ∼16% of individuals were incongruently assigned into species groups with both methods. In contrast, in Robregordo and Montejo, the differences in estimates of individual introgression rates were slight and affected only ∼10% of © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

S G S I N O A K S T A N D S W I T H D I F F E R E N T L E V E L S O F H Y B R I D I Z A T I O N 1213 Robregordo and none of Montejo individuals. In Robregordo, only three individuals (∼3%) were classified into one or the other species, depending on the method. Incongruently assigned individuals (0, 28 and 3, in Montejo, Somosierra and Robregordo, respectively) were removed from further analyses.

Fine-scale spatial genetic structure (SGS) Fine-scale spatial genetic structure in Q. petraea and Q. pyrenaica was low–moderate, with pairwise kinship coefficients only being significantly different from zero in short-distance classes (Fig. 3). Autocorrelograms by species and site suggested different spatial structure patterns for both oak species depending on the population. Density and hybridization rate were both negatively correlated with SGS; i.e. the higher the density and hybridization rate, the lower the amount and range of SGS. In Montejo, where hybrids are rare and trees grow at low densities, SGS in Q. petraea was stronger than in the other sites. Conversely, a higher density and admixture diluted SGS in Somosierra, in particular for Q. petraea. Finally, in Robregordo, where hybridization rates are similar to those found in Montejo (∼6–9%), a five-fold density of sessile oaks, together with heterogeneity in the spatial location of

trees (see Fig. 1b), prevented the formation of significant SGS at intermediate distances. Strength of SGS as shown by pairwise coancestry coefficients at short distances revealed inverted patterns for Q. petraea and Q. pyrenaica in Montejo and Somosierra (Q. pyrenaica had higher shortdistance pairwise coancestry than Q. petraea in Montejo and the opposite in Somosierra; Fig. 3). Nevertheless, both species followed similar trends within each population. For the Montejo and Somosierra populations, it was possible to produce SGS analyses based on both intra and interspecific pairwise kinship coefficients (Fig. 4). Montejo and Somosierra showed contrasting patterns. In Montejo, intraspecific comparisons always gave higher pairwise kinship than interspecific ones, whatever the distance. The slope of the regression of Fij on log-distance for interspecific pairs in Montejo computed along the whole distance range (0–580 m) showed a slightly positive but not significant trend (b-log = 0.0028, SE = 0.0032). The biological reasoning for this trend is not clear, as more distant Q. petraea-Q. pyrenaica pairs seemed to be more related than nearby ones. In contrast, in Somosierra, where the estimate of hybridization rate was high, interspecific pairs showed higher relatedness in the first distance classes (up to 75 m) than at distances above this. The slope of the regression of Fij on log-distance (distance range of 0–830 m) was, in this

Fig. 3 Autocorrelograms of Quercus petraea and Q. pyrenaica in each population, considering: (a) all trees; (b) trees with 0.85 posterior probability of being pure types; and (c) trees with 0.95 posterior probability of being pure types. Circles and triangles represent Q. petraea and Q. pyrenaica individuals, respectively. Values significantly (P < 0.05) different from the expectation under a random distribution of genotypes are indicated by filled symbols. © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

1214 M . V A L B U E N A - C A R A B A Ñ A E T A L .

Fig. 4 Intra- and interspecific autocorrelograms of Quercus petraea and Q. pyrenaica in Montejo and Somosierra mixed populations, considering: (a) pure types jointly with hybrids; and (b) after removing putative hybrids. Circles and triangles represent Q. petraea and Q. pyrenaica individuals, respectively, and squares refer to interspecific pairwise comparisons. Values significantly (P < 0.05) different from the expectation under a random distribution of genotypes are indicated by filled symbols.

case, significantly negative (b-log = −0.0074, SE = 0.0031). An indirect evidence of hybridization can be drawn from this SGS pattern showing a bigger coancestry for interspecific pairs separated by a few meters than pairs separated by larger distances (Fig. 4). Thus, the spatial genetic structure has two different components in Somosierra: an intraspecific component and an interspecific component, which is not present in Montejo apparently. Interestingly, this interspecific component is not present when putative hybrids are removed from the analysis (b-log = 0.0048, SE = 0.0033; Fig. 4b).

Sp statistics The Sp statistic is robust to different sampling schemes, being useful here to compare among species and sites. Sp values by population generally agreed with the inferences from autocorrelograms described above. Q. petraea was found to be consistently more structured than Q. pyrenaica in all sites. However, differences between species from the same population were slight and higher differences were observed among populations than between species. Insights about the effects of hybridization on spatial structure were obtained from spatial analyses for groups of trees with different levels of introgression. For instance, in Montejo the Sp statistic increased only slightly when hybrids were removed from the analyses due to their low numbers in this population (∼6%) and the lack of an

interspecific spatial structure component. In contrast, in Somosierra, there were differences in the Sp statistic across groups of trees with different introgression levels (see Fig. 5). The effect of hybridization on SGS was notable in this population, involving lower pairwise kinship in shorter distance classes but no effects at longer distances (see Figs 3 and 4), which considerably reduced b-log, and might be more relevant for SGS than population differences in density. Indeed, when hybrids were removed from the analyses, populations with five-fold differences in density showed similar Sp values (Montejo and Somosierra). Nevertheless, differences among sites must also be related to other population parameters besides density and hybridization, since Q. petraea in Robregordo, growing at a lower density than in Somosierra and with only a few hybrids, showed the lowest value of Sp found in this study.

Contemporary and historical gene flow in Montejo In previous studies using the Montejo population, a parentage analysis based on naturally regenerated saplings was developed, obtaining direct estimates of contemporary gene flow for this population (Valbuena-Carabaña et al. 2005). Here, these estimates were compared with the historical dispersion parameters obtained from the observed pattern of spatial genetic structure, assuming effective densities (D e) of 0.9, 0.5 and 0.25 times D, the census density. © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

S G S I N O A K S T A N D S W I T H D I F F E R E N T L E V E L S O F H Y B R I D I Z A T I O N 1215 σ−20 σ, at which SGS is shaped by demographical processes under the prediction of the isolation-by-distance model (Hardy et al. 2006).

Discussion Hybridization rates and species discrimination

Fig. 5 Sp statistics in Montejo (squares), Somosierra (rhomboids) and Robregordo (circles) populations, considering: (a) all trees; (b) trees with 0.85 posterior probability of being pure types; and (c) trees with 0.95 posterior probability of being pure types. Confidence intervals at 95% are also given. Notice that the Sp estimate for Q. petraea pure trees in Somosierra was not significantly different from zero due to very large confidence intervals caused by low sample size once hybrids were removed (N = 15).

Contemporary and historical gene flow was similar for Q. petraea in Montejo. Contemporary acorn and pollen axial dispersal standard deviations were 79.33 m and 132.84 m, respectively, as shown by parentage analyses. Then, following Crawford (1984), the combined axial standard deviation of propagule dispersal would be 122.95 m, an estimate very similar to those calculated from SGS for effective densities 0.5–0.9D (σ = 117.04–145.17 m). Thus, if the historical effective densities were truly 0.5–0.9 fold the present-day census density in Montejo, current patterns of gene dispersal in Q. petraea would be similar to the historical ones. In comparison with Q. petraea, parentage analyses showed shorter seed dispersal, σseed = 44.21 m, and longer pollen dispersal, σpollen = 193.45 m, in Q. pyrenaica in Montejo. The combined contemporary axial dispersal standard deviation in this species, σ = 143.76 m, was higher than in Q. petraea but still shorter than the historical estimate (σ = 168.11 m). However, only the De = 0.9D case converged for Q. pyrenaica, and comparisons between contemporary and historical gene flow might not be reliable in this species. Finally, it is worthwhile mentioning that the precision of indirect gene dispersal estimates here might be limited, because the sampled area was much smaller than the range © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

Hybridization in oaks has traditionally been inferred on the basis of morphological traits, an approach favoured by the existence of intermediate hybrid forms (see reviews for Spanish oaks in Vicioso 1950; Franco 1990; Gil et al. 1994). However, hybrid discrimination based on genetic markers in oaks has often provided higher estimates of hybridization rates than morphology (Craft et al. 2002; GonzálezRodríguez et al. 2004; see Valbuena-Carabaña et al. 2005 for Q. petraea-pyrenaica in Montejo). High levels of introgression at cytoplasmic genomes in the absence of intermediate phenotypes have also been reported for several oak species (Whittemore & Schaal 1991; Dumolin-Lapégue et al. 1999; Dodd & Afzal-Rafii 2004). Interspecific controlled crosses have shown large morphological variation in hybrids, including parental and intermediate traits, and presence of extreme or transgressive phenotypes rather than just intermediate ones (e.g. sunflower, Rieseberg & Ellstrand 1993; see a review of 171 hybrid systems in Rieseberg et al. 1999). In the case of the Q. petraea-robur complex, juvenile F1 hybrids exhibited leaf morphologies that were similar to the female parent rather than intermediate (Kleinschmit et al. 1995; Steinhoff 1998). Therefore, morphological traits might not always reflect the genetic admixture of individuals. In this study, morphology and molecular markers provided reasonably similar species discrimination, differing mostly in the estimates of individual admixture rates. Nevertheless, the concordance of morphology and molecular-based species discrimination was higher in the populations with lower hybridization rates (Montejo and Robregordo). Hybridization rates can vary depending on population (e.g. Q. gambelii-grisea, Williams et al. 2001; Q. crassifoliacrassipes, Tovar-Sánchez & Oyama 2004) and mating system (e.g. Q. petraea-robur, Bacilieri et al. 1996; Q. ilex-suber, Belahbib et al. 2001) characteristics. In the populations studied, hybridization rates were not directly related to the species relative abundance, as in Montejo and Robregordo the introgression rates were similar. The spatial segregation of parental species seemed a relevant factor to explain hybridization rates in mixed oak stands. In Montejo, hybrids are only found in a few contact areas, being ecological conditions related to soil depth responsible for the spatial segregation of parental oak species (Pardo et al. 2004; Valbuena-Carabaña et al. 2005). In Somosierra, there are several contact zones between species and individuals introgressed at different levels are more evenly distributed.

1216 M . V A L B U E N A - C A R A B A Ñ A E T A L . Soils in Somosierra are deeper than in Montejo and no differences in species distribution due to soil type are found here. Thus, the higher hybridization rate found in Somosierra might be explained by species proximity due to lack of microenvironmental variation. Indeed, among all feasible scenarios explaining the contrasting hybridization patterns found in Montejo and Somosierra, those related to extrinsic (ecological and population) factors are most likely. Intrinsic forces (related to mating systems and interfertility barriers) probably operate more similarly along the whole geographical range of oak species. Ultimately, the hybridization process depends on different selective pressures on hybrids that might be intrinsic (environmental independent) or extrinsic (see a review in Carney et al. 2000). The establishment and survival of hybrids is greatly affected by their relative fitness compared to the parental species, and on the environmental variation. Theoretical hybridization models that account for environmentally dependent selection of hybrids (e.g. the Hybridization of the Habitat Theory, Anderson 1948; the Bounded Superiority Model, Moore 1977; or the Evolutionary Novelty Model, Arnold 1997) postulate that in nature hybrids are expected to grow in an intermediate zone — the hybrid zone — between the two parental habitats, and that persistence of individuals with lower levels of introgression would require a continuous or jeopardized habitat where various combinations of the two parental habitats were found. Giving these conditions, hybrids would be able to germinate, survive and successfully compete with the parental species. These theories might hold in the Q. petraea-pyrenaica hybrid complex, where extension of the hybrid zone correlated with the population hybridization rate. However, few data are available on environmental variation in hybrid zones and performance of oak hybrids for survival, growth and ecophysiological traits (Himrane et al. 2004 and references therein). Further ecological and quantitative genetics research is necessary to assess selective pressures on Q. petraea and Q. pyrenaica and their hybrids. Hybridization rates in oaks could also depend on opportunities for hybrid formation. For instance, patterns of long distance pollen flow might lead to allopatric hybridization in some populations but not in others (Dodd & Afzal-Rafii 2004). Somosierra and Robregordo are located in the upper lands of a wide and open valley (the Lozoya valley) at both sides of a narrow passage that constitutes a natural mountain pass across the Central System Range (see Fig. 1). At the valley level, a large cloud of Q. pyrenaica’s pollen would be formed a few days before Q. pyrenaica blossoms in Somosierra and Robregordo, situated at higher altitudes. Temperate Q. petraea has, in these populations, an earlier flowering only partly overlapping with Q. pyrenaica’s from the same location. However, phenological overlapping would be enhanced if the above-mentioned pollen cloud,

raising from less elevated woodlands, reached the dehesas studied, increasing interspecific mating opportunities. In contrast, Montejo is sheltered from the Lozoya valley by hills and elevations of marked relief that prevent the access of Q. pyrenaica’s pollen produced at low elevations to Montejo’s Q. petraea (see Fig. 1c).

Intra- and interspecific SGS patterns Greater differences in SGS were observed among populations than between species, although Q. petraea was consistently more structured, in terms of Sp statistics, than Q. pyrenaica at all sites. Species and sites differences in SGS might be related to demographical processes. Simulations showed that a steep decrease of kinship at short distances occurs when gene dispersal follows a leptokurtic distribution (Heuertz et al. 2003; see also Epperson 2005). This decrease can also be originated by differences in the relative contribution of seed and pollen to the overall level of gene dispersal, as is the case in oak species (1:200 seed:pollen dispersal ratios in oaks, Ennos 1994; compared to 1:17, the average in plants, reviewed by Petit et al. 2005). In the light of this study, restricted gene dispersal occurred in both species, the differences in coancestry levels at short distances suggesting less extended acorn dispersal for Q. pyrenaica than for Q. petraea. Previous results based on a parentage analysis in Montejo (Valbuena-Carabaña et al. 2005) and the contrasting average weight of acorns (3.26 ± 0.67 g and 2.14 ± 0.64 g in Q. pyrenaica and Q. petraea, respectively) also support this hypothesis. In temperate forest trees, formation of SGS due to restricted gene dispersal by seed or pollen is common (for instance in oaks, Q. rubra, Sork et al. 1993; Q. laevis, Berg & Hamrick 1995; Q. robur and Q. petraea, Streiff et al. 1998). All oak populations studied showed a significant intraspecific spatial genetic structure. Ecological and demographical factors, particularly density, have a strong influence on fine-scale spatial genetic structure in plants (e.g. aggregation and population geometry, Doligez et al. 1998; Meagher & Vassiliadis 2003; canopy closure, Dutech et al. 2005; secondary dispersal or predation by animals, González-Martínez et al. 2002; Valbuena-Carabaña et al. 2005; see review for breeding systems, density and life forms in Vekemans & Hardy 2004). In the populations studied, SGS was not strongly correlated with present day density. However, present day and historical densities are not necessarily equivalent, with the latter being very difficult to assess. In fact, these dehesas have been intensively exploited for wood, charcoal, acorns and pasture in the past, and oaks might have grown at fluctuating densities due to anthropic pressure. In consonance with this study, a lack of correlation between SGS and population density has been recently reported in tropical tree species (Hardy et al. © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

S G S I N O A K S T A N D S W I T H D I F F E R E N T L E V E L S O F H Y B R I D I Z A T I O N 1217 2006). Other population factors, such as the explicit spatial configuration of the population or the spatial pattern of hybridization, could be more relevant. For instance, the surprisingly low genetic structure for Q. petraea found in Robregordo (Sp = 0.00333, SE = 0.00092; one-fourth Montejo) can be explained on the basis of its particular spatial configuration. Nowadays in Robregordo oaks are found inside dense, albeit patchily distributed, holly tree clumps located along water courses and in humid areas, which provide shade and protection from cattle grazing. Assuming that seed movements were similar to other sessile oak forests, recruitment in this population would be limited to a few suitable patches, thus enlarging effective seed dispersion and reducing SGS. Apart from a significant intraspecific spatial genetic structure within all populations, an interspecific component was also found in Somosierra, a population with high hybridization rates (∼22%). In Somosierra, pairwise interspecific comparisons produced a higher kinship, significantly different from zero, at shorter distance classes (see Fig. 4). On the one hand, an indirect evidence of hybridization and its extent can be inferred from interspecific spatial analysis. Hardy & Vekemans (2001) analysed spatial interactions in Centaurea jacea with different ploidy levels, suggesting the feasibility of this kind of study for any two distinct groups (see also Vekemans & Lefebvre 1997), as theorised by Rousset (1999). On the other hand, insights about the effect of hybridization on SGS patterns can be inferred. Indeed, preferential hybridization at short distances would reduce single species fine-scale genetic structure through lowering kinship among neighbouring plants, simply because the kinship between a pure individual and a hybrid would be smaller than the kinship between two pure individuals of the same species. Preferential hybridization at short distances would result in spatial clustering of introgressed individuals, reducing finescale genetic structure of the species as a whole. To the best of our knowledge, this is the first study reporting the possible effects of hybridization on the formation of fine-scale genetic structure in plants. Although hybridization has often been reported as a process accounting for taxa differentiation (Whittemore & Schaal 1991; Rieseberg 1995), its implications on the differentiation of individuals at the local scale has never been considered before.

Comparison between contemporary and historical gene flow in Montejo Comparisons of contemporary and historical estimates of the variance of axial dispersal, σ2, were made in Montejo, similar to those made by Dutech et al. (2005) for Q. lobata in California. Contemporary (i.e. only for the last established cohort) and historical estimates of gene dispersion were © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

similar. Thus, effective densities of reproductive trees and gene flow patterns have probably not changed much across oak generations in Montejo. Studies comparing contemporary and historical estimates of gene flow are scarce and incomplete because either they lack direct estimates of seed-mediated gene flow (Dutech et al. 2005) or different populations were used for direct and indirect estimates (Ishihama et al. 2005). In order to determine whether method-specific constraints or true demographical variation lead to observed differences in the estimates, more studies are needed, as suggested recently by Hardy et al. (2006).

Concluding remarks As predicted, fine-scale spatial genetic structure of the scattered Q. petraea was consistently stronger than spatial structure found in the more widespread Q. pyrenaica in the same populations, which is also in accordance with specific dispersal capabilities related to acorn weight. Nevertheless, bigger differences in fine-scale genetic structure, which cannot be uniquely attributed to present-day census density, were found among populations than between species. Moreover, for the first time, hybridization has been described as a potentially important factor shaping single species within-population spatial structure due to an excess of nearby interspecific mating and the spatial clustering of resultant hybrids.

Acknowledgements We are grateful to Ricardo Alía for valuable discussion on the original idea of this study. Comments on an early version of the manuscript by J.J. Robledo-Arnuncio and A. Soto are much appreciated. We are indebted to numerous friends and colleagues for leaves collections, especially to B.R. Rubio who also took morphometric measures based on leaves, S. Mariette and C. Burgarella. We thank C. Collada for laboratory technical assistance and P.C. Grant for reviewing the language. María Valbuena-Carabaña was supported by a PhD scholarship from the ‘Comunidad Autónoma de Madrid’ and Santiago C. González-Martínez by a ‘Ramón y Cajal’ fellowship (RC02-2941). This work was funded by CAM 07M/0011/2000 and CAM 07M/0012/2002 projects.

References Anderson E (1948) Hybridization of the habitat. Evolution, 2, 1– 9. Anderson EC, Thompson EA (2002) A model-based method for identifying species hybrids using multilocus genetic data. Genetics, 160, 1217–1229. Arnold ML (1997) Natural Hybridization and Evolution. Oxford University Press, New York. Bacilieri R, Ducousso A, Petit RJ, Kremer A (1996) Mating system and asymmetric hybridization in a mixed stand of European oaks. Evolution, 50, 900–908. Belahbib N, Pemonge M-H, Ouassou A, Sbay H, Kremer Petit RJ (2001) Frequent cytoplasmic exchanges between oak species

1218 M . V A L B U E N A - C A R A B A Ñ A E T A L . that are not closely related: Quercus suber and Q. ilex in Morocco. Molecular Ecology, 10, 2003 –2012. Berg EE, Hamrick JL (1995) Fine-scale genetic structure of a turkey oak forest. Evolution, 49, 110 – 120. Boavida LC, Silva JP, Feijó JA (2001) Sexual reproduction in the cork oak (Quercus suber L). II. Crossing intra– and interspecific barriers. Sexual Plant Reproduction, 14, 143 –152. Burczyk J, Adams WT, Moran GF, Griffin AR (2002) Complex patterns of mating revealed in a Eucalyptus regnans seed orchard using allozyme markers and the neighbourhood model. Molecular Ecology, 11, 2379–2391. Campbell RB (1986) The interdependence of mating structure and inbreeding depression. Theoretical Population Biology, 30, 232– 244. Carney SE, Wolf DE, Rieseberg LH (2000) Hybridisation and forest conservation. In: Forest Conservation Genetics: Principles and Practice (eds Young A, Boshier DH, Boyle TJ. ), pp. 167–182. CSIRO, Melbourne, Australia. Craft KJ, Ashley MV, Koenig WD (2002) Limited hybridization between Quercus lobata and Quercus douglasii (Fagaceae) in a mixed stand in central coastal California. Americal Journal of Botany, 89, 1792–1798. Crawford TJ (1984) The estimation of neighbourhood parameters for plant populations. Heredity, 52, 273 –283. Dodd RS, Afzal-Rafii Z (2004) Selection and dispersal in a multispecies oak hybrid zone. Evolution, 58, 261– 269. Doligez A, Baril C, Joly HI (1998) Fine scale spatial genetic structure with nonuniform distribution of individuals. Genetics, 148, 905–919. Dow BD, Ashley MV (1996) Microsatellite analysis of seed dispersal and parentage of saplings in bur oak, Quercus macrocarpa. Molecular Ecology, 5, 615 – 627. Dow BD, Ashley MV, Howe HF (1995) Characterization of highly variable (GA/CT)n microsatellites in the bur oak, Quercus macrocarpa. Theoretical and Applied Genetics, 91, 137–141. Doyle JJ, Doyle JL (1990) Isolation of plant DNA from fresh tissue. Focus, 12, 13–15. Dumolin-Lapégue S, Kremer A, Petit RJ (1999) Are chloroplast and mitochondrial DNA variation species independent in oaks? Evolution, 53, 1406–1413. Dutech C, Sork VL, Irwin AJ, Smouse PE, Davis FW (2005) Gene flow and fine-scale genetic structure in a wind-pollinated tree species, Quercus lobata (Fagaceaee). American Journal of Botany, 92, 252–261. Endler JA (1977) Geographic Variation, Speciation and Clines. Princeton University Press, Princeton New Jersey. Ennos RA (1994) Estimating the relative rates of pollen and seed migration among plant populations. Heredity, 72, 250 – 259. Epperson BK (1990) Spatial autocorrelation of genotypes under directional selection. Genetics, 124, 757–771. Epperson BK (1993) Recent advances in correlation analysis of spatial patterns of genetic variation. Evolutionary Biology, 27, 95– 155. Epperson BK (2003) Geographical Genetics. Princeton University Press, Princeton New Jersey. Epperson BK (2005) Estimating dispersal from short distance spatial autocorrelation. Heredity, 95, 7–15. Franco J (1990) Quercus. In: Flora Ibérica. Plantas Vasculares de la Península Ibérica e Islas Baleares, Vol. II: 15 –36 (eds Castroviejo S, Lainz M, López-González G, Montserrat P, Muñoz Garmendia F, Paiva J y Villar L). CSIC, Madrid. Gil L, Jiménez MP, Díaz-Fernández PM (1994) Quercus complex in

Spain: an overview of its present state. In: Inter- and Intra-Specific Variation in European Oaks: Evolutionary Implications and Practical Consequences, Brussels, June 15–16, 1994. Office for Official Publications of the European Communities. González Martínez SC, Gerber S, Cervera MT, Martínez-Zapater JM, Gil L, Alía R (2002) Seed gene flow and fine-scale structure in a Mediterranean pine (Pinus pinaster Ait.) using nuclear microsatellite markers. Theoretical and Applied Genetics, 104, 1290–1297. González-Rodríguez A, Arias DM, Valencia S, Oyama K (2004) Morphological and RAPD analysis of hybridization between Quercus affinis and Q. laurina (Fagaceae), two Mexican red oaks. American Journal of Botany, 91, 401–409. Hardy OJ (2003) Estimation of pairwise relatedness between individuals and characterization of isolation-by-distance processes using dominant genetic markers. Molecular Ecology, 12, 1577 – 1588. Hardy OJ, Maggia L, Bandou E, Breyne P, Caron H, Chevallier M-H, Doligez A, Dutech C, Kremer A, Latouche-Hallé C, Troispoux V, Veron V, Degen B (2006) Fine-scale genetic structure and gene dispersal inferences in ten neotropical tree species. Molecular Ecology, 15, 559–571. Hardy OJ, Vekemans X (1999) Isolation by distance in a continuous population: reconciliation between spatial autocorrelation and population genetics models. Heredity, 83, 145–154. Hardy OJ, Vekemans X (2001) Patterns of allozymic variation in diploid and tetraploid Centaurea jacea at different spatial scales. Evolution, 55, 943–954. Hardy OJ, Vekemans X (2002) spagedi: a versatile computer program to analyse spatial genetic structure at the individual or population levels. Molecular Ecology Notes, 2, 618–120. Heuertz M, Vekemans X, Hausman J-F, Palada M, Hardy OJ (2003) Estimating seed versus pollen dispersal from spatial genetic structure in the common ash. Molecular Ecology, 12, 2483 –2495. Himrane H, Camarero JJ, Gil-Pelegrín E (2004) Morphological and ecophysiological variation of the hybrid oak Quercus subpyrenaica (Q. faginea × Q. pubescens). Trees-Structure and Function, 18, 566 – 575. Howard DJ, Britch SC, Braswell WE (2003) Evolution in hybrid zones. Chapter 5. In: The Evolution of Population Biology (eds Singh RK, Uyenoyama MK). Cambridge University Press. Hy FC (1895) Sur quelques chênes hybrides des environs d’Angers. Bulletin de la Société Botanique de France, 42, 552 –559. Ishihama F, Ueno S, Tsumura Y, Washitani I (2005) Gene flow and inbreeding depression inferred from fine-scale genetic structure in an endangered heterostylous perennial, Primula sieboldii. Molecular Ecology, 14, 983–990. Jones AG, Ardren WR (2003) Methods of parentage analysis in natural populations. Molecular Ecology, 12, 2511–2523. Kampfer S, Lexer C, Glössl J, Steinkellner H (1998) Characterization of (GA)n microsatellite loci from Quercus robur. Hereditas, 129, 183–186. Kleinschmit JRG, Bacilieri R, Kremer A, Roloff A (1995) Comparison of morphological traits of pedunculate oak (Quercus robur L.) and sessile oak [Quercus petraea (Matt.) Liebl]. Silvae Genetica, 44, 256–269. Latta RG, Linhart YB, Fleck D, Elliot M (1998) Direct and indirect estimates of seed versus pollen movement within a population of ponderosa pine. Evolution, 52, 61–67. Levin DA (1984) Inbreeding depression and proximity- dependent crossing success in Phlox drummondii. Evolution, 38, 116–127. Linhart YB, Mitton JB, Sturgeon KB, Davis ML (1981) Genetic © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

S G S I N O A K S T A N D S W I T H D I F F E R E N T L E V E L S O F H Y B R I D I Z A T I O N 1219 variation in space and time in a population of ponderosa pine. Heredity, 46, 407–426. Loiselle BA, Sork VL, Nason J, Graham C (1995) Spatial genetic structure of a tropical understorey shrub, Psychotria officinalis (Rubiaceae). American Journal of Botany, 82, 1420–1425. Malécot G (1950) Quelques schémas probabilistes sur la variabilité des populations naturelles. Annales de l’Université de Lyon A, 13, 37–60. Meagher TR, Vassiliadis C (2003) Spatial geometry determines gene flow in plant populations. In: Genes in the Environment (eds Hails R, Beringer J, Godfray HCJ), pp. 76–90. Blackwell Publishing, Oxford. Moore WS (1977) An evaluation of narrow hybrid zones in vertebrates. Quarterly Review of Biology, 52, 263 –277. Pardo F, Gil L, Pardos JA (2004) Structure and composition of pole-stage stands developed in an ancient wood pasture in central Spain. Forestry, 77, 67–74. Petit RJ, Duminil J, Fineschi S, Hampe A, Salvini D, Vendramin GG (2005) Comparative organization of chloroplast, mitochondrial and nuclear diversity in plant populations. Molecular Ecology, 14, 689–701. Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics, 155, 945–959. Rieseberg LH (1995) The role of hybridization in evolution: old wine in new skins. American Journal of Botany, 82, 944 – 953. Rieseberg LH, Ellstrand NC (1993) What can molecular and morphological markers tell us about hybridization? Critical Reviews in Plant Sciences, 12, 213 – 241. Rieseberg LH, Archer MA, Wayne RK (1999) Transgressive segregation, adaptation, and speciation. Heredity, 83, 363–372. Rousset F (1997) Genetic differentiation and estimation of gene flow from F-statistics under isolation by distance. Genetics, 145, 1219–1228. Rousset F (1999) Genetic differentiation within and between two habitats. Genetics, 151, 397–407. Rousset F (2000) Genetic differentiation between individuals. Journal of Evolutionary Biology, 13, 58 – 62. Rousset F (2001) Inference from spatial population genetics. In: Handbook of Statistical Genetics (eds Balding D, Bishop M, Cannings C), pp. 239 – 269. John Wiley & Sons Ltd, Chichester. Rousset F (2004) Genetic Structure and Selection in Subdivided Populations. Princeton University Press, Princeton New Jersey. Smouse PE, Dyer RJ, Westfall RD, Sork VL (2001) Two-generation analysis of pollen flow across a landscape. I. Male gamete heterogeneity among females. Evolution, 55, 260–271. Sork VL, Huang S, Wiener E (1993) Macrogeographic and fine scale genetic structure in a North American oak species, Quercus rubra L. Annales Des Sciences Forestieres, 50 (Suppl.) 1, 261–270. Sork VL, Davis FW, Smouse PE, Apsit VJ, Dyer RJJ, Fernandez M, Kuhn B (2002) Pollen movement in declining populations of California Valley Oak, Quercus lobata: Where have all the fathers gone?. Molecular Ecology, 11, 1657–1668. Stacy EA (2001) Cross-fertility in two tropical tree species: evidence of inbreeding depression within populations and genetic divergence among populations. American Journal of Botany, 88, 1041–1051. Steinhoff S (1998) Controlled crosses between pedunculate and

© 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

sessile oak: results and conclusion. Allgemeine Forst- und Jagdzeitung, 169, 163–168. Steinkellner H, Fluch S, Turetschek E, Lexer C, Streiff R, Kremer A, Burg K, Glössl J (1997) Identification and characterization of (GA/CT)n-microsatellite loci from Quercus petraea. Plant Molecular Biology, 33, 1093–1096. Streiff R, Labbe T, Bacilieri R, Steinkellner H, Glossl J, Kremer A (1998) Within-population genetic structure in Quercus robur L. & Quercus petraea (Matt.) Liebl. assessed with isozymes and microsatellites. Molecular Ecology, 7, 317–328. Tovar-Sánchez E, Oyama K (2004) Natural hybridization and hybrid zones between Quercus crassifolia and Quercus crassipes (Fagaceae) in Mexico: morphological and molecular evidence. American Journal of Botany, 91, 1352–1363. Vähä JP, Primmer CR (2005) Efficiency of model-based Bayesian methods for detecting hybrid individuals under different hybridization scenarios and with different number of loci. Molecular Ecology, 15, 63–72. Valbuena-Carabaña M, González-Martínez S, Sork VL, Collada C, Soto A, Goicoechea PG, Gil L (2005) Gene flow and hybridisation in a mixed oak forest (Quercus pyrenaica Willd. and Quercus petraea (Matts.) Liebl.) in central Spain. Heredity, 95, 457–465. Vekemans X, Hardy OJ (2004) New insights from fine-scale spatial genetic structure analyses in plant populations. Molecular Ecology, 13, 921–935. Vekemans X, Lefebvre C (1997) On the evolution of heavy-metal tolerant populations in Armeria maritima: evidence from allozyme variation and reproductive barriers. Journal of Evolutionary Biology, 10, 175–191. Vicioso C (1950) Revisión del Género Quercus en España. Ministerio de Agricultura, Madrid. Whitlock MC, McCauley DE (1999) Indirect measures of gene flow and migration: FST ≠ (4Nm + 1). Heredity, 82, 117–125. Whittemore AT, Schaal BA (1991) Interespecific gene flow in sympatric oaks. Proceedings of the National Academy of Sciences of USA, 88, 2540–2544. Williams JH, Williams JB, Howard DJ (2001) Reproductive processes in two oak (Quercus) contact zones with different levels of hybridisation. Heredity, 87, 680–690. Wright S (1943) Isolation by distance. Genetics, 28, 114–138.

Maria Valbuena-Carabaña has broad interest in population genetics and conservation of natural forest trees. Her research is focused on genetic structure, gene flow and dispersal being particularly interested in hybridization and introgression processes occurring in nature. Santiago C. González-Martínez is interested in population genetics, adaptation and evolution of Mediterranean plants, in particular forest trees and narrow endemics. Olivier J. Hardy develops methods to infer patterns of gene flow using genetic markers. His research is nowadays focused on tropical tree species with conservation purposes. Luis Gil is senior scientist heading forest genetics research unit at the Forestry College (Polytechnic University of Madrid). His lab focuses on population genetics, breeding and conservation of forest genetic resources, in particular for Iberian conifers, oaks and elms.