- Split View
-
Views
-
Cite
Cite
COLIN T. KELLEHER, TREVOR R. HODKINSON, GERRY C. DOUGLAS, DANIEL L. KELLY, Species Distinction in Irish Populations of Quercus petraea and Q. robur: Morphological versus Molecular Analyses, Annals of Botany, Volume 96, Issue 7, December 2005, Pages 1237–1246, https://doi.org/10.1093/aob/mci275
- Share Icon Share
Abstract
• Background and Aims Populations of oak (Quercus petraea and Q. robur) were investigated using morphological and molecular (AFLP) analyses to assess species distinction. The study aimed to describe species distinction in Irish oak populations and to situate this in a European context.
• Methods Populations were sampled from across the range of the island of Ireland. Leaf morphological characters were analysed through clustering and ordination methods. Putative neutral molecular markers (AFLPs) were used to analyse the molecular variation. Cluster and ordination analyses were also performed on the AFLP markers in addition to calculations of genetic diversity and F-statisitcs.
• Key Results A notable divergence was uncovered between the morphological and molecular analyses. The morphological analysis clearly differentiated individuals into their respective species, whereas the molecular analysis did not. Twenty species-specific AFLP markers were observed from 123 plants in 24 populations but none of these was species-diagnostic. Principal Coordinate Analysis of the AFLP data revealed a clustering, across the first two axes, of individuals according to population rather than according to species. High FST values calculated from AFLP markers also indicated population differentiation (FST = 0·271). Species differentiation accounted for only 13 % of the variation in diversity compared with population differentiation, which accounted for 27 %.
• Conclusions The results show that neutral molecular variation is partitioned more strongly between populations than between species. Although this could indicate that the populations of Q. petraea and Q. robur studied may not be distinct species at a molecular level, it is proposed that the difficulty in distinguishing the species in Irish oak populations using AFLP markers is due to population differentiation masking species differences. This could result from non-random mating in small, fragmented woodland populations. Hybridization and introgression between the species could also have a significant role.
INTRODUCTION
There has been much debate on species concepts in European oak, in particular for Quercus petraea (Matt.) Liebl. and Q. robur L. The debate has ranged from difficulties with species separation based on morphological traits (Stace, 1975) to suggestions of alternative species concepts to accommodate such an ‘awkward’ taxonomic group (Van Valen, 1976). It has been suggested that the biological species concept is less appropriate for oak (Burger, 1975) owing to levels of gene exchange and hybridization within the genus. However, although there are difficulties involved, the species Q. petraea and Q. robur have been differentiated into relatively distinct entities using morphological analyses (Dupouey and Badeau, 1993; Aas, 1995; Kleinschmit et al., 1996; Bruschi et al., 2000).
With the continued use of molecular techniques for describing diversity and species characteristics, a potential source of markers is available for characterizing oaks and for testing species differentiation at the molecular level. However, molecular markers have generally been found to be less useful than morphological characters in differentiating between Q. petraea and Q. robur, and very few studies have shown a distinction between the species based on molecular data (Scotti-Saintagne et al., 2004). There are no significant differences in genome size between the species (Zoldos et al., 1998). A study of multiple-copy ribosomal genes in the two species found conserved genome organization (Zoldos et al., 1999). No species-diagnostic markers have been found from molecular analyses, the only differences identified thus far being some frequency differences in loci abundances. In a detailed study of molecular markers, only 2 % of 2800 PCR products obtained from RAPD primers showed significant frequency differences between the species (Moreau et al., 1994; Bodénès et al., 1997). The difficulty in distinguishing the species at the molecular level and obtaining species-diagnostic markers is possibly due to high levels of intraspecific diversity and considerable hybridization between the species (Rushton, 1993). This hybridization and introgression results in the boundaries of species delimitation being less definitive. However, two studies to date have shown a species distinction at the molecular level. In a study by Muir et al. (2000) a distinction between European populations of Q. petraea and Q. robur was shown based on an analysis of the proportion of shared microsatellite alleles. A distinction between the species has also been observed using AFLP in some Belgian oak populations (Coart et al., 2002). Ongoing studies on European oak are increasing our understanding of the morphological/molecular duality in these closely related species.
The aim of the research presented here was to investigate the species distinction of populations of oak by comparing morphological and molecular analyses and to situate Irish populations in a European context. In a previous study, a morphological distinction of oak species in Irish populations was shown to be clearly evident (Kelleher et al., 2004b). In this paper the results are presented from the analysis of AFLP markers (Vos et al., 1995) to compare the differentiation found at the molecular level. AFLP has been shown to be useful for discriminating between closely related species of other angiosperms (Hodkinson et al., 2000, 2002). Thus, this method was chosen to try to elucidate some of the species and population characteristics of Irish oak species.
MATERIALS AND METHODS
Sampling
We sampled broadly across the island of Ireland to obtain a representative sample of both oak species. Sites were sampled from 24 geographical locations across Ireland (Fig. 1). Sites were chosen primarily as putative ancient woodland sites. Some Irish sites have been provisionally identified as ancient woodland using historical records, floristics and an assessment of woodland structure (Kelly and Fuller, 1988; Rackham, 1995; Bohan, 1997). Where evidence was lacking, sites were chosen for their remoteness, thus lessening the likelihood of in-planting of external stock. Sites were predominantly monospecific, except in a few instances where in-planting was known to have occurred. Site details are presented elsewhere (Kelleher, 2001; Kelleher et al., 2004b). Five trees were sampled per site, except at four sites (see legend to Fig. 1). Leaf samples were collected from open-grown canopy branches. This minimized variation in leaf morphology resulting from environmental factors such as exposure and aspect (Baranski, 1975; Blue and Jensen, 1988). Five well-developed leaves per tree were collected and analysed morphologically to determine species status. Leaf samples were labelled and pressed. Leaves for the molecular analysis were collected and stored in dried silica gel until DNA was extracted.
Morphological analysis
The morphological data were analysed as previously described in Kelleher et al. (2004b). Leaves were assessed for 16 variables: nine measurements and seven derived characters. The morphological dataset was converted to z-scores and a distance matrix was prepared based on the Euclidean distance. The resulting distance matrix was used for a Principal Coordinate Analysis (PCoA) using R Package version 4.0 d1 software (Casgrain et al., 1999). The distance matrix was also used in the ‘Neighbour-Joining’ (NJ) cluster analysis (Saitou and Nei, 1987). The NJ tree of the morphological data was used to designate a species grouping for each individual sample (Kelleher et al., 2004b).
AFLP analysis
DNA was extracted from the leaves using a modified hot CTAB protocol (Doyle and Doyle, 1987). AFLP reactions were carried out using the Applied Biosystems AFLP™ Plant Mapping kit (Applied Biosystems, Warrington, UK). The kit contained Mse1 and EcoR1 ligation adaptor pairs, pre-selective primers, selective primers and Core Mix (composed of a mix of buffer, nucleotides and AmpliTaq® DNA polymerase). The reaction protocol was as suggested in the kit, except that the reaction volumes were halved.
Each restriction-ligation reaction contained 250 ng DNA, 1× Promega T4 Ligase buffer, 0·05 m NaCl, 0·275 mg BSA, 0·5 U Mse1, 2·5 U EcoR1, 0·5 U Promega T4 DNA Ligase, 0·5 μL Mse1 Adaptor Pair, 0·5 μL EcoR1 Adaptor and nanopure water up to a volume of 5·5 μL. The restriction-ligation reactions were carried out at 37 °C for 2·5 h in a PTC 100 MJ Research Thermal Cycler. On completion, 94·5 μL of 1× TE0·1 (20 mm Tris–HCl, 0·1 mm EDTA, pH 8·0) was added to the reactions and they were stored at −20 °C until used.
A pre-selective amplification was carried out using the pre-selective primers supplied with the kit. The reactions contained 2·0 μL diluted restriction-ligation product, 8·0 μL AFLP Pre-selective Primers and 7·5 μL Core Mix. The PCRs were run on an PTC −200 MJ Research thermal cycler with the following reaction specifications: an initial extension at 72 °C for 2 min, followed by 20 cycles of 94 °C for 1 s, 56 °C for 30 s and 72 °C for 2 min, followed by a final extension at 60 °C for 30 min. A 4 °C soak ended the reaction. All ramp times were set at 1 °C s−1. Five microlitres of the PCR product was diluted in 95 μL TE0·1 and the reactions were stored on ice or at 2–6 °C until used.
A selective amplification reaction was performed for each primer pair with each reaction containing: 1·5 μL diluted PCR pre-selective amplification product, 0·5 μL EcoR1-labelled primer, 0·5 μL Mse1 primer and 7·5 μL Core Mix. A number of selective amplification trials were run and the primers used in the final analysis were: EcoR1-ACA and Mse1-CAC (FAM labelled), EcoR1-ACG and Mse1-CTC (JOE labelled), and EcoR1-AAC and Mse1-CTG (NED labelled). The selective amplification PCR was a drop-down PCR starting with 2 min of denaturing at 94 °C, and the reaction cycled as: 94 °C for 1 s, x °C for 30 s and 72 °C for 2 min (where x was decreased 1 °C at a time for each cycle from 65 to 56 °C, which continued for 23 cycles). The final extension was at 60 °C for 30 min and a soak at 4 °C stopped the reactions. All ramp times were set to 1 °C s−1.
The three PCR products for each sample were multiplexed as follows: 0·6 μL FAM product, 0·8 μL JOE product, 1·3 μL NED product, 0·5 μL ROX 500S (Applied Biosystems) size standard and 24 μL formamide (Applied Biosystems). This was heated to 95 °C for 5 min and then stored on ice until use. The samples were then run on an ABI PRISM® 310 Genetic Analyser. Electropherograms were read and fragments sized using GeneScan® Analysis Software Version 3.1 and then processed using the Genotyper® Software Version 3.7 (Applied Biosystems).
AFLP data handling and analysis
Ordination and clustering
The resulting distance matrix was used for analysis of the AFLP data through PCoA using the R Package version 4.0 d1 (Casgrain et al., 1999) and NJ clustering using PAUP 4* (Swofford, 1999).
Chi-squares were calculated manually to test for significant frequency differences of the fragments between the respective species.
Genetic diversity and differentiation
Gene diversity and levels of genetic differentiation were estimated using AFLP-SURV version 1.0 (Vekemans et al., 2002). Values of total diversity, within-population diversity, between-population diversity and population differentiation were calculated. The populations were analysed first as a whole group (i.e. data for Q. petraea and Q. robur combined) and second were partitioned into the separate species. To estimate levels of differentiation between the species, the species were grouped and analysed as two separate populations corresponding to the morphological species designations. Hardy–Weinburg equilibrium was assumed for all calculations. Statistical nomenclature followed Vekemans et al. (2002), where HW is analogous to HS, and HB is the average diversity between groups/populations. Significance of the FST values was calculated based on comparison with values obtained from randomly permuted individuals among the populations. In addition, to test the findings from the FST statistics an Analysis of Molecular Variance (AMOVA) was carried out using Arlequin version 2.001 (Schneider et al., 2000). The data were analysed using a three-level hierarchical analysis in which variation was partitioned between species groups, among populations of the species groups and within the individuals of each population.
RESULTS
Species designation
Species designations were based on the procedure implemented previously in Kelleher et al. (2004b) using NJ analysis of the leaf morphological data (Fig. 2). One population (AR—Ballymascanlan) used for the AFLP analysis was collected when in bud and the species was determined after leaf emergence. The morphological groupings of Q. petraea and Q. robur were defined by splitting the NJ tree at the greatest point of divergence i.e. mid-point rooting (Fig. 2). Suspected hybrids or introgressed individuals were allocated to the taxon to which they appeared closest in the NJ analysis. An overlay of the NJ designations on a PCoA of the morphological data reveals the agreement with the groupings established from each method (Fig. 3). From 123 individuals, 94 were designated as Q. petraea and 29 as Q. robur.
AFLP fragments
AFLP profiles were scored and analysed for 123 samples from 24 populations across Ireland. A total of 147 fragments obtained from three primer sets were used in the analysis. On average, each fragment was shared by 35 individuals, although the distribution of the fragments was skewed with many fragments having very few occurrences (Fig. 4). A large proportion of the fragments occurred in only 1–10 individuals (Fig. 4). There were seven fragments common to all samples (leaving a total of 140 polymorphic fragments). Twenty species-specific fragments were observed, although these corresponded to the 20 population-specific markers and many of these (17) were individual-specific (Table 1). No species-diagnostic fragments (fragments consistently in one species and not in the other) were observed, although four fragments showed significant deviation from the expected chi-squared values (values for the four fragments were: χ2 = 7·55, 8·87, 7·27, P ≤ 0·01; and 14·54, P ≤ 0·001; d.f. = 1).
Population . | No. of population-specific markers . | No. of individual-specific markers . |
---|---|---|
Abbey Leix | 4 | 3 |
Charleville | 9 | 7 |
Crolly | 1 | 1 |
Curraghmore | 1 | 1 |
Gearagh | 1 | 1 |
Cappercullen Glen | 3 | 3 |
Derryclare | 1 | 1 |
Totals | 20 | 17 |
Population . | No. of population-specific markers . | No. of individual-specific markers . |
---|---|---|
Abbey Leix | 4 | 3 |
Charleville | 9 | 7 |
Crolly | 1 | 1 |
Curraghmore | 1 | 1 |
Gearagh | 1 | 1 |
Cappercullen Glen | 3 | 3 |
Derryclare | 1 | 1 |
Totals | 20 | 17 |
Population . | No. of population-specific markers . | No. of individual-specific markers . |
---|---|---|
Abbey Leix | 4 | 3 |
Charleville | 9 | 7 |
Crolly | 1 | 1 |
Curraghmore | 1 | 1 |
Gearagh | 1 | 1 |
Cappercullen Glen | 3 | 3 |
Derryclare | 1 | 1 |
Totals | 20 | 17 |
Population . | No. of population-specific markers . | No. of individual-specific markers . |
---|---|---|
Abbey Leix | 4 | 3 |
Charleville | 9 | 7 |
Crolly | 1 | 1 |
Curraghmore | 1 | 1 |
Gearagh | 1 | 1 |
Cappercullen Glen | 3 | 3 |
Derryclare | 1 | 1 |
Totals | 20 | 17 |
Principal Coordinate Analysis
From a PCoA of the morphological data, a species distinction is clearly evident (Fig. 3). However, the groupings obtained with the PCoA of the AFLP data (Fig. 5) were not congruent with those obtained by morphological analyses (Fig. 3). The AFLP markers were not able to separate the species decisively on the basis of the first two principal axes (Fig. 5). The PCoA of the AFLP data revealed that just over 40 % of the overall variation was accounted for by the first ten axes—the first three axes accounted for 15, 7 and 3 %, respectively. Contrary to the morphological analyses (see groupings in Fig. 2) a grouping of many individuals into their respective populations was evident from the AFLP analysis (Fig. 6). Axis 3 did not enhance the resolution of the species distinction (data not shown). When other axes were investigated, in particular axes 4 and 5, a species split was more evident (Fig. 7). Thus, the emergent pattern from the PCoA of the AFLP analysis is that of population groupings rather than species groupings.
Population and species genetic diversity
Gene diversity values are shown in Table 2. Diversity levels were similar to those from other studies using dominant markers. The average within-population diversity (HS = 0·1626) was higher than the average between-population diversity (HB = 0·0604) (Table 2). Population differentiation was greater when all populations were analysed (FST = 0·271) than when the species groups were analysed (FST = 0·1304). Population differentiation, FST, is most evident in Q. petraea populations and with the populations taken as a whole, whereas it is less marked in Q. robur and with differentiation between the species (Table 2). The diversity statistics show a similar pattern to that obtained from the PCoA, i.e. population structure is more pronounced than species structure in the populations analysed. The AMOVA also shows that a greater amount of the overall variation is attributable to within-population variation (63 %) than to between-population variation (38 %) and that the variation accounted for by the species groups is negligible (Table 3).
Populations . | HT . | HS . | s.e. (HS) . | HB . | FST . |
---|---|---|---|---|---|
All (n = 123) | 0·2230 | 0·1626 | 0·0054 | 0·0604 | 0·2710* |
Species (n = 123) | 0·2147 | 0·1867 | 0·00435 | 0·0280 | 0·1304* |
Q. petraea (n = 94) | 0·2437 | 0·1774 | 0·005284 | 0·0664 | 0·2726* |
Q. robur (n = 29) | 0·2645 | 0·2301 | 0·01682 | 0·0344 | 0·1289* |
Populations . | HT . | HS . | s.e. (HS) . | HB . | FST . |
---|---|---|---|---|---|
All (n = 123) | 0·2230 | 0·1626 | 0·0054 | 0·0604 | 0·2710* |
Species (n = 123) | 0·2147 | 0·1867 | 0·00435 | 0·0280 | 0·1304* |
Q. petraea (n = 94) | 0·2437 | 0·1774 | 0·005284 | 0·0664 | 0·2726* |
Q. robur (n = 29) | 0·2645 | 0·2301 | 0·01682 | 0·0344 | 0·1289* |
Total population diversity, HT; average within-population diversity, HS; average diversity between populations, HB; the population differentiation, FST. n = the number of individuals used in each analysis. Calculations were made by using the program AFLP-SURV version 1.0.
Significant at P < 0·0001.
Populations . | HT . | HS . | s.e. (HS) . | HB . | FST . |
---|---|---|---|---|---|
All (n = 123) | 0·2230 | 0·1626 | 0·0054 | 0·0604 | 0·2710* |
Species (n = 123) | 0·2147 | 0·1867 | 0·00435 | 0·0280 | 0·1304* |
Q. petraea (n = 94) | 0·2437 | 0·1774 | 0·005284 | 0·0664 | 0·2726* |
Q. robur (n = 29) | 0·2645 | 0·2301 | 0·01682 | 0·0344 | 0·1289* |
Populations . | HT . | HS . | s.e. (HS) . | HB . | FST . |
---|---|---|---|---|---|
All (n = 123) | 0·2230 | 0·1626 | 0·0054 | 0·0604 | 0·2710* |
Species (n = 123) | 0·2147 | 0·1867 | 0·00435 | 0·0280 | 0·1304* |
Q. petraea (n = 94) | 0·2437 | 0·1774 | 0·005284 | 0·0664 | 0·2726* |
Q. robur (n = 29) | 0·2645 | 0·2301 | 0·01682 | 0·0344 | 0·1289* |
Total population diversity, HT; average within-population diversity, HS; average diversity between populations, HB; the population differentiation, FST. n = the number of individuals used in each analysis. Calculations were made by using the program AFLP-SURV version 1.0.
Significant at P < 0·0001.
Source of variation . | d.f. . | Sum of squares . | Variance component . | Percentage of variation . |
---|---|---|---|---|
Among species | 1 | 42·3 | −0·34 (NS) | −2·1 |
Among populations within species | 22 | 904·8 | 6·19 (*) | 38·59 |
Within populations | 99 | 1008·6 | 10·18 (*) | 63·51 |
Totals | 122 | 1955·7 | 16·04 |
Source of variation . | d.f. . | Sum of squares . | Variance component . | Percentage of variation . |
---|---|---|---|---|
Among species | 1 | 42·3 | −0·34 (NS) | −2·1 |
Among populations within species | 22 | 904·8 | 6·19 (*) | 38·59 |
Within populations | 99 | 1008·6 | 10·18 (*) | 63·51 |
Totals | 122 | 1955·7 | 16·04 |
Significant at P < 0·001; NS, not significant.
Source of variation . | d.f. . | Sum of squares . | Variance component . | Percentage of variation . |
---|---|---|---|---|
Among species | 1 | 42·3 | −0·34 (NS) | −2·1 |
Among populations within species | 22 | 904·8 | 6·19 (*) | 38·59 |
Within populations | 99 | 1008·6 | 10·18 (*) | 63·51 |
Totals | 122 | 1955·7 | 16·04 |
Source of variation . | d.f. . | Sum of squares . | Variance component . | Percentage of variation . |
---|---|---|---|---|
Among species | 1 | 42·3 | −0·34 (NS) | −2·1 |
Among populations within species | 22 | 904·8 | 6·19 (*) | 38·59 |
Within populations | 99 | 1008·6 | 10·18 (*) | 63·51 |
Totals | 122 | 1955·7 | 16·04 |
Significant at P < 0·001; NS, not significant.
DISCUSSION
Diversity and population structure
As oak is predominantly an outbreeding species it is expected that it would have considerable genetic diversity. The levels of diversity in the Irish populations (HT = 0·223) were of similar magnitude but lower than those obtained in Flemish populations (Table 2; HT in the Flemish populations ranged from 0·29 to 0·30, Coart et al., 2002). These values were also similar to those obtained from Q. petraea populations analysed using other dominant markers (Le Corre et al., 1997, HT = 0·24). Quercus robur populations showed greater average diversity than Q. petraea (Table 2), although the opposite has been shown in other European populations (Bodénès et al., 1997). Population differentiation (FST = 0·271) was much greater than in the Flemish populations (FST = 0·02–0·07). There is a possible upward bias in the FST values for the overall populations as a result of lower sample sizes (Nei and Chesser, 1983); however, this higher level of population differentiation was also evident in cpDNA from the same Irish populations (Kelleher et al., 2004a). Comparing a number of studies on cpDNA from Iberia, France, Britain and Ireland it was also evident that the level of diversity reduces from south to north (Kelleher et al., 2004a). Although organellar markers will show a greater degree of differentiation compared with nuclear markers owing to their strictly maternal inheritance and dispersal through seeds, they still indicate a similar trend to those observed with AFLP markers. In a recent study of Irish populations of Q. petraea, using nuclear SSR markers, it was proposed that there is low population differentiation in Ireland (Muir et al., 2004). Their study, however, was restricted to one species and was based on only seven sites and so does not represent the entire island. From the present study and our prior work (Kelleher et al., 2004a), both encompassing an island-wide range of sites, we find clear evidence of population differentiation across Ireland. We suggest that this population differentiation is a factor that is obscuring the species distinction at the molecular level. The reduction in genetic diversity and increased population differentiation in the Irish populations is suggestive of inbreeding effects, which are not uncommon in island populations (Frankham, 1997).
Contrasting patterns in morphological and molecular analyses
The morphological analysis reveals a high degree of separation between the species (Fig. 3). Based on the PCoA of the first two axes, 60 % of the overall variation in the morphological dataset can be attributed to species differences. This sharply contrasts with the results of the AFLP PCoA. Based on the most discriminating axes for the species distinction, approximately 6 % of the overall variation in the AFLP data can be attributed to species differences (Fig. 7). A chi-squared analysis of the AFLP markers also showed no evidence for significant frequency differences in the markers in the respective species. A greater proportion of the variation can be attributed to differences between populations (Fig. 6). The dominant pattern obtained from the AFLP analysis is a grouping of many individuals into their respective populations. By contrast, individuals did not group into their respective populations in the morphological analysis (Fig. 2). The population groupings from the AFLP analysis hold no evident geographical explanation, as populations from widely separated localities are just as likely to cluster close together as are those from adjoining localities (Fig. 6). An NJ tree of the AFLP data based on genetic similarity (Sørensen–Dice measure) yielded the same conclusion (data not shown). This suggests possible random genetic drift within the populations.
The PCoA showing the population clusters (Fig. 6) can be explained by the levels of population differentiation discussed above, i.e. population differentiation results in individuals of the same population clustering together. To estimate levels of differentiation in the populations and the species, the data were partitioned and analysed as population groups (24 populations) and species groups (two species: Q. petraea and Q. robur). Comparing these levels of differentiation, it is evident that the population grouping shows greater differentiation than the species grouping (FST, Table 2). The variability due to the species differentiation (HB, Table 2) is approximately 13 %, whereas that for population differentiation is 27 %. This suggests that at a molecular level variation is partitioned more between populations than between species, with little evidence for species differentiation. In a previous study on these populations using cpDNA markers, no species distinction was evident from the cpDNA haplotypes, although some differences in haplotype distribution between the species were found (Kelleher et al., 2004a).
The pattern obtained from the Irish populations differs with that obtained from Flemish populations, which were shown to differentiate by species on the basis of AFLP markers (Coart et al., 2002). Even so, based on the first two axes of their PCoA, the species differences accounted for only 13·5 % of the total variation in the Flemish populations. Species differences in Irish populations could be masked by reduced diversity coupled with non-random mating, in addition to gene exchange through hybridization and introgression. Hybridization and introgression have already been indicated from morphological analysis in Irish populations of oak (Kelleher et al., 2004b). Because the AFLP analysis produces many fragments (most of which occur in only a few individuals; Fig. 4), it is possible that the multivariate analysis was sensitive to a relatively small number of markers co-occurring in populations. The co-occurrence of markers would be a result of increased levels of inbreeding and genetic drift. Inbreeding and genetic drift effects are highly likely in small, isolated and fragmented populations, such as those in Ireland. Less than 1 % of the land area in Ireland is under native woodland cover and the majority of these areas are small woods separated by farmland (based on information from the Forest Inventory and Planning System Unit, Department of the Marine and Natural Resources, Ireland, 2000). The present study supports the evidence for limited differentiation of Q. petraea and Q. robur at the molecular level as found in other studies (Bodénès et al., 1997; Zoldos et al., 1999). The data could also be interpreted as showing evidence against the species status of the two species involved, but this is hard to justify when viewed from the morphological perspective. It is more likely that the differentiation at the molecular level is at a limited number of adaptive loci—hence the clear morphological distinction but less apparent distinction using neutral loci such as AFLP. In a recent study assessing genomic regions differentiating the two species, it was found that coding regions expressed higher differentiation than non-coding regions (Scotti-Saintagne et al., 2004). This is reflected in the morphological markers (coding regions) showing differentiation, whereas the neutral markers (non-coding) lack that definitive differentiation. The findings presented here highlight the problem in differentiating the species and emphasize the importance of combining both morphological and molecular analyses. Although levels of inbreeding are implied from the results, it would be of interest to assess whether this is affecting fitness through genetic load. As oak is a highly variable, predominantly outbreeding species, it would probably be many generations before fitness is affected.
CONCLUSIONS
A comparison of morphological and molecular analyses in Irish oak populations shows that the species Q. petraea and Q. robur are much more clearly distinguished using morphological traits than using neutral molecular markers (AFLPs). The species distinction was shown to represent a lesser element in the overall variation based on the AFLP markers and population differentiation was more pronounced. It is proposed that the difficulty in distinguishing the species in Irish oak populations using AFLP markers is due to population differentiation masking the species differences. This could result from non-random mating in small, fragmented woodland populations. Hybridization and introgression between the species could also have played a significant role.
We thank all site owners who allowed us access to their land for sampling. We also thank our funding body, COFORD, the National Council for Forest Research and Development.
LITERATURE CITED
Aas G.
Baranski M.
Blue MP, Jensen RJ.
Bodénès C, Joandet S, Laigret F, Kremer A.
Bohan R.
Bruschi P, Vendramin GG, Bussotti F, Grossoni P.
Casgrain P, Legendre P, Vaudor A.
Coart E, Lamote V, De Loose M, Van Bockstaele E, Lootens P, Roldán-Ruiz I.
Doyle JJ, Doyle JL.
Dupouey JL, Badeau V.
Frankham R.
Hodkinson TR, Chase MW, Takahashi C, Leitch IJ, Bennett MD, Renvoize SA.
Hodkinson TR, Renvoize SA, Ní Chonghaile G, Stapleton CMA, Chase MW.
Kelleher CT.
Kelleher CT, Hodkinson TR, Douglas GC, Kelly DL.
Kelleher CT, Hodkinson TR, Kelly DL.
Kelly DL, Fuller S.
Kleinschmit J, Elsner G, Schlums K.
Le Corre V, Dumolin-Lapègue S, Kremer A.
Moreau F, Kleinschmit J, Kremer A.
Muir G, Lowe AJ, Fleming CC, Vogl C.
Nei M, Chesser RK.
Rackham O.
Rushton BS.
Saitou N, Nei M.
Schneider S, Roessli D, Excoffier L.
Scotti-Saintagne C, Mariette S, Porth I, Goicoechea PG, Barreneche T, Bodenes K, et al.
Swofford DL.
Vekemans X, Beauwens T, Lemaire M, Roldan-Ruiz I.
Vos P, Hogers R, Bleeker M, Reijans M, van de Lee T, Hornes M, et al.
Zoldos V, Papes D, Brown SC, Panaud O, Siljak-Yakovlev S.
Author notes
1Department of Botany, Trinity College, School of Natural Sciences, University of Dublin, Dublin 2, Ireland and 2Teagasc, Kinsealy Research and Development Centre, Malahide Road, Dublin 17, Ireland