| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Population Biology |
2Department of Botany, North Carolina State University, Raleigh, North Carolina 27695-7612 USA 3Department of Ecology and Evolution, University of Chicago, Chicago, Illinois 60637 USA 4Department of Ecology and Evolutionary Biology and the Natural History Museum and Biodiversity Research Center, University of Kansas, Lawrence, Kansas 66045-2106 USA; 5Department of Biology and Huck Institutes of Life Sciences, Pennsylvania State University, University Park, Pennsylvania 16802 USA
Received for publication June 1, 2005. Accepted for publication October 24, 2005.
ABSTRACT
Knowledge regarding the origin and maintenance of hybrid zones is critical for understanding the evolutionary outcomes of natural hybridization. To evaluate the contribution of historical contact vs. long-distance gene flow in the formation of a broad hybrid zone in central and northern Georgia that involves Aesculus pavia, A. sylvatica, and A. flava, three cpDNA regions (matK, trnD-trnT, and trnH-trnK) were analyzed. The maternal inheritance of cpDNA in Aesculus was confirmed via sequencing of matK from progeny of controlled crosses. Restriction site analyses identified 21 unique haplotypes among 248 individuals representing 29 populations from parental species and hybrids. Haplotypes were sequenced for all cpDNA regions. Restriction site and sequence data were subjected to phylogeographic and population genetic analyses. Considerable cpDNA variation was detected in the hybrid zone, as well as ancestral cpDNA polymorphism; furthermore, the distribution of haplotypes indicates limited interpopulation gene flow via seeds. The genealogy and structure of genetic variation further support the historical presence of A. pavia in the Piedmont, although they are at present locally extinct. In conjunction with previous allozyme studies, the cpDNA data suggest that the hybrid zone originated through historical local gene flow, yet is maintained by periodic long-distance pollen dispersal.
Key Words: Aesculus • cpDNA inheritance • hybrid zone • phylogeography • Pleistocene • Sapindaceae • secondary contact • southeastern United States
Hybrid zones provide a natural setting for the observation of evolutionary processes, such as adaptation, speciation, and introgression (Arnold, 1997
; Avise, 2004
). Although hybridization was once regarded by some as simply "evolutionary noise" (e.g., Wagner, 1969
, 1970
), recent views consider hybridization and hybrid zones as vehicles for the creation of novel species and adaptations, as well as mechanisms for augmenting genetic diversity, and for either strengthening or reducing reproductive isolating barriers between species (e.g., Rieseberg and Wendel, 1993
; Arnold, 1997
; Rieseberg, 1997
). Hybridization and introgression have highly influenced the distribution, diversity, and occurrence of species observed today, given their evolutionary significance and prevalence (Stace, 1987
; Rieseberg and Wendel, 1993
; Ellstrand et al., 1996
). Awareness of the genetic structure and introgression within hybrid zones and those species involved in their formation is critical to predicting the potential evolutionary outcomes of natural hybridization. Therefore, a necessary step toward this goal is determining the genetic composition of hybrid zones via detailed multilocus analyses involving independent loci from both nuclear and cytoplasmic genomes (McCauley, 1995
; Ouborg et al., 1999
; Avise, 2004
). Traditional population genetics often interprets genetic similarity between populations as a result of gene flow. However, genetic similarity between taxa may also be attributable to shared common ancestry. Thus, it is important to combine traditional population genetic analyses with a phylogeographic approach (e.g., Avise et al., 1987
; Schaal et al., 1998
; Hewitt, 2001
) to determine the relative contributions of shared ancestry and genetic exchange among species in hybrid zones. Although the phylogeographic approach is typically applied to the study of intraspecific variation, in a complex of hybridizing species, taxonomic barriers often become indistinct, rendering the phylogeographic approach a useful tool for analyzing the geographic patterns of genetic variation within a complex of recently diverged species (e.g., García-Paris et al., 2000
; Beheregay et al., 2004
). In a hybrid zone, alleles (or haplotypes) shared between parental species can be a result of gene introgression via interspecific gene flow, recency of common ancestry of the hybridizing species, or a combination of both of these phenomena. Furthermore, alleles shared between parental species and hybrid populations may be caused by both historical and current gene flow that can be both local and long distance. Although distinguishing these alternative processes is not an easy task, it is essential to the understanding of hybrid zone evolution and may be achieved via synthesis of evidence from gene genealogy and patterns of genetic variation of loci that differentially track the histories of dispersal. In flowering plants, pollen and seeds are the typical dispersal agents, with the latter more vagile than the former in most plants.
In the genus Aesculus L. (Sapindaceae), the seeds are large and heavy (Schopmeyer, 1974
), thus it is not likely that seeds will be distributed across broad geographic ranges. In contrast, the pollen may be dispersed long distances via the ruby-throated hummingbird (dePamphilis and Wyatt, 1989
). This discrepancy between dispersal distances of pollen and seed allows for the elucidation of the relative contribution of pollen vs. seeds to the genetic architecture of the hybrid zone via a combination of loci from nuclear and cpDNA genomes. Nuclear markers will be distributed by both pollen and seed, while only seeds, given the maternal inheritance of chloroplasts, will disperse cpDNA markers.
The genus Aesculus is comprised of trees and shrubs that are widely cultivated for their showy, large inflorescences and dense foliage. The genus consists of 13–19 species discontinuously distributed in four separate regions: eastern Asia, Europe, western North America, and eastern North America (Hardin, 1957a
; Fang, 1981
). Four species of the genus in eastern North America constitute sect. Pavia (Mill.) Person. Hybridization among three species of sect. Pavia, A. pavia L., A. flava Ait., and A. sylvatica Bartr., has resulted in the creation of a hybrid zone in central to northern Georgia and adjacent areas (Hardin, 1957b
; dePamphilis and Wyatt, 1989
). This hybrid zone is notable for its broad width (~200 km) and geographic and genetic asymmetry. Hybrids between A. pavia and A. sylvatica occur within the geographic range of A. sylvatica, but nearly 145 km from the nearest populations of A. pavia (Hardin, 1957b
; dePamphilis and Wyatt, 1989
, 1990
). The natural distribution of A. sylvatica is limited to the deciduous and pine forests of the Piedmont from the southern limits of Virginia to northwestern Alabama, while A. pavia occurs in the mixed pine and deciduous forests of the southern Coastal Plain; A. flava is distributed throughout deciduous forests in the Appalachian Mountain range and northward to the Ohio River Valley (Hardin, 1957c
; Fig. 1). To explain the asymmetrical distribution of hybrids, Hardin (1957b
, 1957c
) proposed that species of sect. Pavia were once allopatric in the Appalachians and evolved divergently and that hybridization occurred during a period of secondary contact as a result of range expansion of A. pavia during the interglacial warm periods of the Pleistocene. The present-day absence of A. pavia from the hybrid zone is posited to be a result of southward range reduction of the species from the Piedmont and subsequent localized extinction of A. pavia in the Piedmont.
|
), the climatic oscillations of the Pleistocene led to the creation of refugia during periods of cooler temperatures (Hewitt, 1996
, 2000
, 2001
). Populations that survived in refugia migrated and expanded their ranges during periods of warmer climates, resulting in a subsequent reduction in genetic diversity in the newly colonized regions. As a consequence of the range expansions and contractions of newly diverged species, hybrid zones lacking one or both parental species could be formed (Remington, 1968
; Hewitt, 1996
) because hybrids were able to survive in environments that were no longer suitable for one or both of the parental species. Thus, the extinction of A. pavia in the hybrid zone could have been caused by environmental selection favoring hybrids and discriminating against A. pavia parentals in the Piedmont region of northern Georgia. Remington (1968)
identified several areas, or suture zones, where range expansions and contractions were likely to bring together previously allopatric species. One such zone is located in the southeastern United States.
dePamphilis and Wyatt (1989
, 1990
) conducted an allozyme and pollination biology study of Aesculus in the hybrid zone. Their allozyme data were concordant with the morphological data, with both confirming the broad width and genetic asymmetry of the hybrid zone. Although there were no fixed species-specific alleles, computer simulations and admixture analyses were used to provide evidence that multilocus genotypes and overall population structures were indicative of hybridization, in agreement with earlier morphological studies. dePamphilis and Wyatt (1989)
posited that the genetic presence of A. pavia in the hybrid zone reflects long-distance pollen dispersal of A. pavia by ruby-throated hummingbirds (Archilochus colubris). The ruby-throated hummingbird is a common pollinator of all three Aesculus species, and it migrates from the Coastal Plain to the Piedmont each year during the flowering season of Aesculus (dePamphilis and Wyatt, 1989
).
The two proposed hypotheses, historical secondary contact (HSC) vs. recurrent long-distance gene flow (LDF), may be distinguished by the current genetic composition of hybrid populations. If hybridization occurred during the Pleistocene, followed by local extinction of A. pavia, the presence of A. pavia and A. flava cpDNA haplotypes in the Piedmont and hybrid zone would provide strong evidence for the hypothesis of HSC between presently allopatric species. In contrast, hybridization by recent long-distance pollen dispersal should result in some individuals in hybrid populations containing mostly A. pavia nuclear genes, while chloroplast genes of A. pavia (if inherited maternally) should be extremely rare or absent in the zone, given the probable limited dispersal distance of the large and heavy Aesculus seeds (Schopmeyer, 1974
). Hybrid populations containing nuclear markers predominantly from one species and the cytotype of another species would reveal the nature of the contribution of each species to the hybrid zone, whether it be through pollen dispersal, seed dispersal, or both. Thus, evidence from the cpDNA genome, when compared to the previously published allozyme analysis of dePamphilis and Wyatt (1989
, 1990
) will allow us to further assess the HSC hypothesis. In the current study, we present data from three regions of cpDNA in order to examine and confirm the mechanisms that have contributed to the origin and maintenance of the hybrid zone.
Although cpDNA is generally inherited maternally in angiosperms (Harris and Ingram, 1991
; Birky, 2001
), instances of nonmaternal inheritance have been documented (e.g., Sewell et al., 1993
). Therefore, it is critical to first verify the mode of cpDNA inheritance in Aesculus so that observed geographic patterns of cpDNA haplotypes may then be attributed to seed dispersal. Thus, our objectives were to (1) determine the mode of cpDNA inheritance in Aesculus, (2) examine the pattern of cpDNA variation in the hybrid zone, and (3) compare results of the present cpDNA analysis to the nuclear allozyme analysis of dePamphilis and Wyatt (1989
, 1990
) for the purpose of inferring the evolutionary history of the Aesculus hybrid zone.
MATERIALS AND METHODS
Sampling and DNA extraction
Leaf samples of A. pavia, A. sylvatica, A. flava, and hybrids between A. pavia and A. sylvatica and A. flava and A. sylvatica were collected from 29 natural populations, both within and outside the boundaries of the hybrid zone (Fig. 1). Species and putative hybrids were identified based on the diagnostic features of Hardin (1957c)
. Three to 26 individuals were sampled per population, with most population samples consisting of 10 individuals (Table 1). A total of 248 individuals were analyzed. Samples were collected from both "pure" allopatric populations of A. flava, A. pavia, and A. sylvatica and from populations sympatric and parapatric to the hybrid zone (Fig. 1). DNA was extracted from fresh or silica-dried leaf materials using the mini-prep method of Cullings (1992)
, with modifications described in Xiang et al. (1998)
. Additionally, fresh leaf material was obtained from parents and F1 progeny of controlled crosses of Aesculus spp. for 17 crosses, with one F1 analyzed per cross (Table 2), from crosses performed at the Holden Arboretum (Kirtland, Ohio, USA).
|
|
. Amplification was carried out on a Peltier Thermal Cycler (PTC-100, Bio-Rad, Hercules, California, USA) with a 50-µl reaction volume, using approximately 80 ng template DNA (3 µl), 5 µl 10x Mg-free buffer, 6 µl 25 mM MgCl2, 8 µl 2.5 mM each dNTP, 2.0 µl 10 mg/ml BSA, 0.3 µl Taq DNA polymerase (5 U/µl), 0.5 µl 20 µM matK1F (forward) and matK3R (reverse) primer, and 24.7 µl sterile deionized water. Hot-start PCR conditions were as follows: an initial 6 min at 96°C, followed by the addition of Taq DNA polymerase, 30 cycles of 40 s at 94°C, 1 min at 45°C, and 2 min 30 s at 72°C, followed by a final cycle of 40 s at 94°C, 1 min at 45°C, and 4 min at 72°C. Polymerase chain reaction products were purified by precipitation using 20% PEG (polyethylene glycol)/2.5 mol NaCl/L sterile deionized water, as done in Morgan and Soltis (1993)
. Purified PCR products were sequenced with the ABI Prism dRhodamine Terminator Cycle Sequencing Ready Reaction kit (Applied Biosystems, Foster City, California, USA) on a PTC-100 following the procedure of Fan and Xiang (2003)
. Sequences were aligned visually, and nucleotide site mutations and indels were determined visually as well. Polymorphic sites between progeny and parentals were visually inspected on chromatographs for errors in base calling. A total of 1372 base pairs (bp), spanning the majority of the matK gene, were compared.
cpDNA restriction site variation
Three cpDNA regions were analyzed for all populations: the matK gene (1830 bp), the trnD (tRNA-Val-GUC)–trnT (tRNA-Gly-GGU) (1415 bp) intergenic spacer, and the trnH (tRNA-Val-GUG)–trnK (tRNA-Phe-UUU) intergenic spacer (2024 bp). Amplification of the matK gene was performed using primers matK1F (forward) and matK1R (reverse) developed by Sang et al. (1997)
, as described. Amplification of the trnD-trnT and trnH-trnK intergenic spacers was performed with primers trnD-F and trnT-R for the trnD-trnT region and trnH-F and trnK-R for the trnH-trnK region (Demesure et al., 1995
). PCRs were conducted in a 50-µl reaction volume as described, except with an annealing temperature of 50°C for trnD-trnT and 63°C for trnH-trnK.
PCR products of a subset of representative samples were subjected to restriction site analysis with 10 enzymes for all regions (TaqI, HhaI, MseI, RsaI, Hsp92II, MspI, HindIII, EcoRI, BamHI, and BsrBI) and two additional enzymes for the matK gene (BsrI and BstNI) to screen for species-specific markers. Based on the results of the screening, BsrI, MspI, BstNI, and TaqI were used for matK, RsaI and MseI for the trnD-trnT region, and BsrBI for the trnH-trnK region. DNA samples from all populations were then PCR-amplified, and restriction digests were performed based on screening results. Digestion was performed with 20-µl reaction volumes containing 5–15 µl of PCR products for 3 to 3.5 h of incubation at enzyme specific temperatures, in separate reactions for each enzyme. The restriction digests were subsequently electrophoresed on a 0.7% agarose/0.7% synergel gel for 3.5 to 4 h at 100 V. Gels were stained with ethidium bromide and digitally photographed using a UV transilluminator and imaging system (Kodak Digital Science 1D, Rochester, New York, USA). The cpDNA haplotype for each sample was determined based on the combined chloroplast banding pattern at the three loci.
Each restriction site haplotype was subsequently sequenced for all regions. All sequences used in this study, including those used for determining the inheritance of chloroplasts in Aesculus, were submitted to GenBank (accession nos. AY968606–AY68671, Appendix). For the sequencing of matK, the primers matK3F, matK3R, and matK2R (Sang et al., 1997
) were used in addition to matK1F and matK1R. For trnD-trnT, the primers of Demesure et al. (1995)
as well as two internal primers designed for Aesculus, trnDT-F2 (5'-TGCCTCCTTGAAAGAGAGATG-3') and trnDT-R2 (5'-CCGTTCGCAGATTTTCAGAT-3') were used. For sequencing of trnH-trnK, the primers of Demesure et al. (1995)
and three internal primers designed for Aesculus, trnH2F (5'-ACTCGTATACACGAAGATCG-3'), trnH3F (5'-CTTATAGCCCCGTGTCAACC-3'), and trnK2R (5'-TGAACCCGTTTCTGGATCTC-3') were used. Aesculus glabra, the fourth species of the monophyletic sect. Pavia (Xiang et al., 1998
), was included in the study to serve as the outgroup for rooting of the haplotype genealogy.
|
), with some manual modifications. The aligned sequences were utilized in nucleotide diversity analysis and genealogy reconstruction. Genetic diversity was estimated with DnaSP 4.00 (Rozas et al., 2003
) for each locus. Indices calculated include the total number of mutations (Eta), the average number of nucleotide substitutions between two sequences per site (
, Nei, 1987
), and the average number of nucleotide differences between two sequences (
, Tajima, 1983
). The genealogy of haplotypes was estimated using two approaches, phylogenetic analysis and construction of a minimum spanning tree. Modeltest 3.6 (Posada and Crandall, 1998
) was used to determine the model that best fit the evolution of the concatenated sequence data. The neighbor-joining (NJ) algorithm (Saitou and Nei, 1987
) implemented in PAUP* (version 4.04b10; Swofford, 2003
) was used to construct a tree based on the maximum likelihood distance method with the general-time-reversible (GTR) model (Rodríguez et al., 1990
) indicated to best fit the data according to Modeltest results. Support for the tree was estimated using bootstrap analysis with 10 000 replicates (Felsenstein, 1985
). The Bayesian analysis was conducted with the program MrBayes v. 3.0 (Huelsenbeck and Ronquist, 2001
), while burn-in was determined using the program Tracer version 1.1 (Jamin and Lautenbaucher, 1993
). The Bayesian analysis was run for 100 000 generations, well past the convergence point of 5700 generations, with trees sampled and saved every 100 generations. The trees were loaded into PAUP* (Swofford, 2003
) to construct a majority rule consensus tree after removing the first 57 trees estimated prior to the convergence point. Frequency values of the trees serve as estimates of the posterior probability of nodes (Huelsenbeck and Ronquist, 2001
). A minimum spanning tree (Excoffier and Smouse, 1994
) was reconstructed for the haplotype sequences using the program MINSPNET packaged within Arlequin ver. 2.000 (Schneider et al., 2000
) to compare with the genealogy derived from phylogenetic analysis.
Population genetic analyses
A data matrix of band presence/absence data from restriction site patterns was constructed to estimate F statistics and to examine the molecular variation using Arlequin ver. 2.000 following Schneider et al. (2000)
. FST values estimated by Arlequin were used to perform an analysis of molecular variance (AMOVA) (Weir and Cockerham, 1984
; Excoffier et al., 1992
; Weir, 1996
). Several grouping strategies were implemented for the purpose of exploring the partitioning of genetic variation, with the first set including three groups, separated by physiogeographic region: the Coastal Plain, Piedmont, and Appalachian Mountains. For the second approach, the Piedmont was further divided into two groups, the Piedmont and the hybrid zone, with all other groups remaining the same. To test whether variation was partitioned among species, AMOVA was performed using both the parental species and hybrid taxa as groups, as well as with only the parental species as groups. Individual analyses for each parental and hybrid taxa were also performed separately, but these tests were lacking in statistical power, due to low sample size. To test for the effects of non-independence of restriction site bands from different enzymes, AMOVA was also performed with data from individual enzymes. The results did not differ significantly from the AMOVA results from data combined from all enzymes, thus are not presented here. Pairwise estimates of FST and the number of migrants per generation, Nm, (Slatkin, 1985
) were estimated with Arlequin ver. 2.000 (Schneider et al., 2000
) with 10 000 permutations and a significance level of 
= 0.05. Estimates of Nm were calculated according to the derivation of Nm for haploid genomes (Birky et al., 1989
), where Nm = 0.5(1/(FST–1)).
RESULTS
Haplotype distribution and diversity
Restriction site analysis of the three loci, matK, trnD-trnT, and trnH-trnK identified 21 unique haplotypes (Appendix S1, see Supplemental Data accompanying online version of this article). A haplotype was designated as unique if the combined restriction fragment-banding pattern from all three loci was distinguishable from every other haplotype. In other words, each haplotype had at least one restriction site or length polymorphism distinguishing it from all other haplotypes. Seven haplotypes (5, 6, 8, 9, 14, 15, 20) were identified in A. pavia, with six of them detected only in that species (5, 6, 9, 14, 15, 20). Four haplotypes (7, 8, 16, 19) were detected in A. sylvatica, of which three were apparently unique to the species (7, 16, 19). Five haplotypes (1, 3, 4, 10, 18) were found in A. flava, all of which were A. flava specific. Additionally, one haplotype (21) was found in only one A. pavia x sylvatica population (SP3), and five (2, 11, 12, 13, 17) were found only in A. sylvatica x flava populations (SF1–3). Two of the A. pavia-specific haplotypes (6, 14) occur in hybrid populations between A. pavia and A. sylvatica (SP1, SP5, and SP7), while no A. flava- or A. sylvatica-specific haplotypes were found in any hybrid populations. One haplotype (8) was common among A. pavia, A. sylvatica, and A. pavia x sylvatica hybrids, and widespread in the hybrid zone and surrounding areas (Fig. 2). Haplotypes 1, 3, 4, 17, and 18 were restricted to the Appalachian Mountains, while other haplotypes (i.e., 5, 7, 9, 10, 11, 12, 13, 15, 16, 19, 20, and 21) were unique to single populations throughout the Piedmont and Coastal Plain. Five haplotypes (1, 4, 6, 8, and 14) occurred in more than one population. Most populations are fixed for a single haplotype, with the exception of the six polymorphic populations, F3, F4, P2, S1, SF1, and SP7.
|
|
). Because the minimum spanning tree is also designed to incorporate allele frequencies and population structure, the rarity of haplotype 2, which is found in two individuals in one population, is most likely affecting its placement on the minimum spanning tree. Nonetheless, the overall lack of spatial and taxonomic structure can be seen in both estimates of genealogical relationships. The minimum spanning tree suggests that three of the sampled haplotypes are ancestral (8, 6, 11) based on their internal positions (Castello and Templeton, 1994
). Haplotype 8 is positioned internally and is connected to more haplotypes than other internal haplotypes on the minimum spanning tree (Fig. 4). This haplotype is also widely distributed among populations (Fig. 2) and potentially represents the most ancestral haplotype of those surveyed.
|
|
|
|
= 0.05 level. Pairwise FST estimates, along with the estimates of Nm (Table 6) revealed that the number of migrants per generation is greater than that would be expected (Nm = 1) if allele frequencies were governed solely by random genetic drift (Wright, 1931
). Furthermore, more migration events seem to be occurring or have occurred between A. sylvatica and hybrid populations than between A. pavia and hybrid populations, or between A. flava and hybrid populations. This is expected given the close geographic proximity of A. sylvatica populations to hybrid populations.
|
Evidence of ancestral chloroplast polymorphism and of historical secondary contact via seeds
Results from the controlled hybridization study (Appendix S2, see Supplemental Data accompanying online version of this article) confirm the maternal inheritance of cpDNA in Aesculus. This evidence permits us to determine the maternal parent of hybrid populations and to trace the pattern of seed flow. Given the purported limited dispersal distance of Aesculus seeds as contrasted with the potential for long-distance pollen dispersal, we expect to see a genealogy and genetic structure in which haplotypes cluster according to species or geographic regions and a pattern of genetic variation showing significant population structure. This pattern was not seen; the polymorphisms do not cluster according to taxonomic groups or geographic regions (i.e., Appalachian Mountains, Piedmont, and Coastal Plain) (Figs. 3, 4; Tables 4, 5). This lack of structure suggests either ancestral cpDNA polymorphism in sect. Pavia (e.g., divergent chloroplast DNA haplotypes within species were inherited from a polymorphic common ancestor of the species in sect. Pavia), or extensive ancient cpDNA capture among species. The ancestral polymorphism hypothesis seems more plausible based on congruencies of the species phylogeny (Fig. 3b) to clades I and II of the cpDNA genealogy (Fig. 3a). Within clades I and II (Fig. 3a), it appears that A. sylvatica diverged first, followed by A. pavia and then A. flava. This branching order is consistent with that found in the ITS/morphology phylogenies (Fig. 3b) constructed by Xiang et al. (1998)
and Forest et al. (2001)
. The branching order of clade III is not clearly consistent with the ITS/morphology species phylogeny. However, based on the presence of three major clades, each containing all three species of Aesculus in the NJ tree (Fig. 3a), it is reasonable to conclude that at least three different cytotypes were present in the most recent common ancestor of A. sylvatica, A. pavia, and A. flava. Chloroplast capture would have occurred randomly and thus concordance of plastid and ITS phylogenies is not expected. Additionally, invoking a hybridization scenario to explain the present geographical distribution of haplotypes necessitates the assumption of extensive ancient gene flow via seeds across a broad geographic scale. Otherwise, it is difficult to account for the grouping of widely disjunct populations from different species into a single clade (Fig. 3a; P5-F8-S5; P4-F1-S2). The results of the AMOVA analysis indicate that, while there is little differentiation at the broadest spatial scale (i.e., Appalachian Mountains, Piedmont, and Coastal Plain), populations within these regions have a high degree of genetic isolation. This is congruent with the expectation of limited seed dispersal and ancestral polymorphism of cpDNA in each species. The isolation of cpDNA among populations is further corroborated in the distributional pattern of haplotypes; most populations are fixed for one haplotype, and most haplotypes are not shared among populations (Fig. 2, Table 1). The lack of spatial structure across species (Fig. 2, Table 5) and physiogeographic provinces (Table 4) further confirms that lineage sorting and ancestral chloroplast polymorphism are affecting the pattern and structure of variation in the zone. The combined results of the AMOVA and genealogy further support the occurrence of ancestral polymorphism of cpDNA and limited seed dispersal on a broad spatial scale (i.e., across physiogeographic provinces) as well as on a less extensive, between-population scale. Ancestral polymorphism of cpDNA has resulted in an inflated estimate of the migration rates of seeds (Table 6), which is derived from FST values. The inflated estimate illustrates an inherent limitation of F statistics, which cannot distinguish between gene flow and shared common ancestry because they are based solely on genetic similarities and do not take into consideration population history. In the case of the Aesculus hybrid zone, the occurrence of particular cytotypes among species and different spatial regions is likely a result of gene flow, while the sharing of closely related, but not identical, cytotypes among species and regions is the product of shared common ancestry.
Under the HSC hypothesis, we expect to observe that some A. pavia x sylvatica populations share haplotypes with A. pavia or have cpDNA haplotypes grouped with A. pavia in the genealogy. The sharing of haplotype 6 between A. pavia (P2) and A. pavia x sylvatica populations (SP1 and SP5) and haplotype 14 between A. pavia (P6) and A. pavia x sylvatica (SP7) (Fig. 3a, clade III) strongly supports the HSC hypothesis and further implies that A. pavia was the maternal parent of these hybrid populations. This evidence indicates that A. pavia was historically present in the hybrid zone and had contact with A. sylvatica. Admixture analysis of allozyme loci (dePamphilis and Wyatt, 1990
) indicated that the nuclear genome of the hybrid populations SP1 and SP5 is composed mainly of A. sylvatica alleles, suggesting that while A. pavia was the maternal parent of these populations, the A. pavia nuclear genes in these hybrid populations have been diluted through backcrosses to A. sylvatica by way of localized pollen dispersal.
Haplotype 8 is shared by most hybrid A. sylvatica x pavia populations (SP 2, 4, 6, 7); it is also found in populations P1 and S1, 3 and 4. Thus the maternal parent of these hybrid populations may be A. pavia, A. sylvatica, or both. Given that haplotype 8 is closely related to haplotype 6 and haplotype 14 of A. pavia, it appears likely that haplotype 8 originated from A. pavia but that the cpDNA donor of haplotype 8 to the hybrid populations most likely varies, depending on the distance of hybrid populations to potential donor populations. Allozyme data (dePamphilis and Wyatt, 1990
) suggests a contribution of both A. pavia and A. sylvatica to the nuclear genome of hybrid individuals in these populations. Additionally, the widespread distribution of haplotype 8 in the hybrid zone suggests that it may be favored by exogenous environmental conditions in the Piedmont.
The remaining hybrid population (SP3) is closely related to A. sylvatica (S5) (Fig. 3a), suggesting that A. sylvatica is the cpDNA donor to this population. However, a large proportion (0.676) of SP3 contains A. pavia nuclear alleles (dePamphilis and Wyatt, 1990
), suggesting that A. pavia has contributed significantly to this population via long-distance pollen dispersal.
A possible alternative explanation to secondary contact for the aforementioned observations of shared chloroplast haplotypes between species, of the ancestral positions of hybrid haplotypes (h8, 11, Fig. 4) and for the presence of each species on all major clades in the NJ tree is that the hybrid zone is a zone of primary intergradation of a single species currently in the process of speciation. Primary intergradation would occur as the zone was formed in response to selection across continuous populations of the species prior to speciation (Harrison, 1990
, 1993
). Evidence from morphology, hybrid fitness values, and DNA sequence divergence between the species strongly supports the secondary contact hypothesis and refutes the primary intergradation hypothesis. The combination of morphological features distinguishing the four species (Hardin, 1957c
; dePamphilis and Wyatt, 1989
) and evidence of lower relative fitness in the form of reduced pollen viability for hybrids in contrast to the "pure species" (dePamphilis and Wyatt, 1989
) offer compelling testimony for secondary contact rather than primary intergradation. According to a calibrated ITS clock of 1.722 + –0.21 x 10–9 nucleotide substitutions per site per year (Xiang et al., 1998
), A. flava and A. sylvatica diverged about 10.2 mya, while A. flava and A. pavia diverged ~5.1 mya, suggesting that they had diverged long before the climatic oscillations of the Pleistocene (~1.8 mya).
Possible long-distance downstream seed dispersal
dePamphilis and Wyatt (1989)
theorized that the genetic presence of A. flava in the hybrid zone was due to dispersal of seeds downstream. However, the terrain of the Appalachian Mountains and the flow of rivers in the southeastern United States (Fig. 2) must also be taken into account. Aesculus flava seeds that fall on the western side of the mountain range will flow west-southwest into the Mississippi River and downstream into the Gulf of Mexico, thus bypassing the central and eastern portions of the hybrid zone. Seeds falling on the eastern side of the mountain will fall into rivers such as the Nantahala that flow into central and eastern portions of the hybrid zone. However, populations situated farther north in North Carolina would be most likely to drop their seeds into rivers flowing into the Atlantic Ocean, rather than through the hybrid zone. While it is possible that gene flow by pollen dispersal may occur between populations of A. flava situated throughout the mountains, claims of gene flow via seeds are unsubstantiated. The hybrid A. sylvatica x flava population (SF1) in the hybrid zone is not closely related to those in the mountains (Fig. 3a), suggesting that long-distance pollen dispersal on a broad geographic scale can not be verified from the results of this study. In the dePamphilis and Wyatt (1989)
study, populations of A. flava sampled in northern Georgia could have contributed to the hybrid zone through dispersal of seeds downstream because they are situated on the eastern slope of the Appalachian Mountains. In our study, all populations of A. flava sampled, with the exception discussed next, were situated on the westward side of the Appalachian Mountains. In contrast, any close affinity of A. flava haplotypes with those in the central and eastern parts of the hybrid zone or with A. sylvatica and A. pavia detected in our study may be due to the past sympatry of these species, most likely in the Piedmont. One possible exception to this generalization is the population of A. flava in Rabun County in northern Georgia, where its seeds could fall into rivers in the Nantahala watershed. In accordance with this argument, one hybrid specific haplotype (11 in SF1) is located at an internal position on the haplotype network, giving rise to the haplotypes found in a "pure" A. flava population (10 in F1, Fig. 4). This suggests that the donor population of this haplotype was once present in the hybrid zone and has become extinct or was not sampled, or that seed dispersal has occurred across a more limited geographic scale.
Conclusions
Our analysis of genealogical relationships, spatial distribution of cpDNA haplotypes, measures of diversity, and partitioning of cpDNA variation has enabled us to address key issues regarding the origin and present structure of the Aesculus hybrid zone. Our results suggest ancestral polymorphism of the chloroplast genome of Aesculus sect. Pavia and abundant cpDNA diversity in the Aesculus hybrid zone. A small number of haplotypes observed in hybrid populations may no longer be present in parental populations. The observed patterns of genetic variation and sharing of chloroplast haplotypes between A. pavia and hybrid populations support Hardin's (1957c) hypothesis of historical secondary contact between A. pavia and A. sylvatica in the Piedmont. In addition to the evidence from allozyme studies (dePamphilis and Wyatt, 1989
, 1990
), which support the long-distance dispersal hypothesis, the cumulative evidence supports the idea that secondary contact of the species during the Pleistocene, followed by both local gene flow from A. sylvatica and long-distance gene flow via A. pavia pollen, were the formative and maintaining mechanisms for the Aesculus hybrid zone. Although it is not uncommon for hybrid zones to have been formed via secondary contact in the Pleistocene (Barton and Hewitt, 1985
), evidence of mechanisms for their stability is often lacking (Harrison, 1990
). If ruby-throated hummingbirds are in fact responsible for the long-distance pollen dispersal events, it is also possible that the ongoing annual hummingbird migration continues to be an active force in maintaining and perhaps even in expanding the hybrid zone in Aesculus and other plant species in the eastern United States. Results from our study suggest that the relationship between vertebrate migration patterns and the genetic structure of plant species should be investigated further to gain insight into the forces that shape the genetic architecture of hybrid zones.
FOOTNOTES
1 The authors thank C. Chan of the Holden Arboretum and S. Wiegrefe of the Morton Arboretum for materials used in the cpDNA inheritance analysis. We thank T. Lasseigne and the J. C. Raulston Arboretum for allowing us to perform initial crossing experiments. The study was supported by a Faculty Research grant from Idaho State University and a Faculty Research and Development grant from NCSU, both awarded to J. Q. Y. X. 
6 Author for correspondence (e-mail: jenny_xiang{at}ncsu.edu
; phone: 919-515-2728; fax: 919-515-3436
LITERATURE CITED
Arnold M. L. 1997. Natural hybridization and evolution Oxford University Press Oxford, UK.
Avise J. C. 2004. Molecular markers, natural history, and evolution, 2nd ed Sinauer Sunderland, Massachusetts, USA.
Avise J. C Arnold J Ball R. M Bermingham E Lamb T Neigel J. E Reeb C. A Saunders N. C. 1987. Intraspecific phylogeography: the mitochondrial DNA bridge between population genetics and systematics. Annual Review of Ecology and Systematics 18: 489-522.[ISI]
Barton N. H Hewitt G. M. 1985. Analysis of hybrid zones. Annual Review of Ecology and Systematics 16: 113-148.[CrossRef][ISI]
Beheregay L. B Gibbs J. P Havill N Fritts T. H Powell J. R Caccone A. 2004. Giant tortoises are not so slow: rapid diversification and biogeographic consensus in the Galápagos. Proceedings of the National Academy of Sciences, USA 101: 6514-6519.
Birky C. W. 2001. The inheritance of genes in mitochondria and chloroplasts: laws, mechanisms, and models. Annual Review of Genetics 35: 125-148.[CrossRef][ISI][Medline]
Birky C. W Fuerst P Maruyama T. 1989. Organelle gene diversity under migration, mutation and drift: equilibrium expectations, approach to equilibrium, effects of heteroplasmic cells, and comparisons to nuclear genes. Genetics 121: 613-627.
Castello J Templeton A. R. 1994. Root probabilities for intraspecific gene trees under neutral coalescent theory. Molecular Phylogenetics and Evolution 3: 102-113.[CrossRef][Medline]
Cullings K. W. 1992. Design and testing of a plant-specific PCR primer for ecological and evolutionary studies. Molecular Ecology 1: 233-240.
Demesure B Sodzi N Petit R. J. 1995. A set of universal primers for amplification of polymorphic non-coding regions of mitochondrial and chloroplast DNA in plants. Molecular Ecology 4: 129-131.[Medline]
dePamphilis C. W Wyatt R. 1989. Hybridization and introgression in buckeyes (Aesculus: Hippocastanaceae): a review of the evidence and a hypothesis to explain long-distance gene flow. Systematic Botany 14: 593-611.[CrossRef][ISI]
dePamphilis C. W Wyatt R. 1990. Electrophoretic confirmation of interspecific hybridization in Aesculus (Hippocastanaceae) and the genetic structure of a broad hybrid zone. Evolution 44: 1295-1317.[CrossRef][ISI]
Ellstrand N. C Whitkus R Rieseberg L. H. 1996. Distribution of spontaneous plant hybrids. Proceedings of the National Academy of Sciences, USA 93: 5090-5093.
Excoffier L Smouse P. E Quattro J. M. 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131: 479-491.[Abstract]
Excoffier L Smouse P. E. 1994. Using allele frequencies and geographic subdivision to reconstruct gene trees within a species: molecular variance parsimony. Genetics 136: 343-359.[Abstract]
Fan C Xiang J. Q.-Y. 2003. Phylogenetic analyses of Cornales based on 26S rRNA and combined 26S rDNA-matK-rbcL sequence data. American Journal of Botany 90: 1357-1372.
Fang W. P. 1981. Hippocastanaceae. In W. P. Fang [ed.], Flora republicae popularis sinicae, 274–289. Science Press, Beijing, China.
Felsenstein J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783-791.[CrossRef][ISI]
Forest F Drouin J. N Charest R Brouillet L Bruneau A. 2001. A morphological phylogenetic analysis of Aesculus L. and Billia Peyr. (Sapindaceae). Canadian Journal of Botany 79: 154-169.[ISI]
García-Paris M Good D. A Parra-Olea G Wake D. B. 2000. Biodiversity of Costa Rican salamanders: implications of high levels of genetic differentiation and phylogeographic structure for species formation. Proceedings of the National Academy of Sciences, USA 97: 1640-1647.
Hardin J. W. 1957a. A revision of the American Hippocastanaceae, II. Brittonia 9: 173-195.[CrossRef]
Hardin J. W. 1957b. Studies in the Hippocastanaceae. III. A hybrid swarm in the buckeyes. Rhodora 59: 45-51.
Hardin J. W. 1957c. Studies in the Hippocastanaceae. IV. Hybridization in Aesculus. Rhodora 59: 185-203.
Harris S. A Ingram R. 1991. Chloroplast DNA and biosystematics: the effects of intraspecific diversity and plastid transmission. Taxon 40: 393-412.
Harrison R. G. 1990. Hybrid zones: windows on evolutionary processes. In J. Antonovics and D. Futuyama [eds.], Oxford surveys in evolutionary biology, 71–128. Oxford University Press, Oxford, UK.
Harrison R. G. 1993. Hybrids and hybrid zones: historical perspective. In R. G. Harrison [ed.], Hybrid zones and the evolutionary process, 3–12. Oxford University Press, Oxford, UK.
Hewitt G. M. 1993. After the ice: Parallelus meets Erythropus in the Pyrenees. In R. G. Harrison [ed.], Hybrid zones and the evolutionary process, 140–164. Oxford University Press, Oxford, UK.
Hewitt G. M. 1996. Some genetic consequences of ice ages, and their role in divergence and speciation. Biological Journal of the Linnean Society 58: 247-276.[CrossRef]
Hewitt G. M. 2000. The genetic legacy of the Quaternary ice ages. Nature 405: 907-913.[CrossRef][Medline]
Hewitt G. M. 2001. Speciation, hybrid zones and phylogeography—or seeing genes in space and time. Molecular Ecology 10: 537-549.[CrossRef][Medline]
Huelsenbeck J Ronquist F. 2001. MrBayes: Bayesian inference of phylogenetic trees. Bioinformatics 17: 754-755.
Jamin M Lautenbacher M. E. 1993. TRACER, version 1.1: a Mathematica package for gamma-algebra in arbitrary dimensions. Computer Physics Communications 74: 265-268.[CrossRef]
McCauley D. E. 1995. The use of chloroplast DNA polymorphism in studies of gene flow in plants. Trends in Ecology and Evolution 10: 198-202.
Morgan D. R Soltis D. E. 1993. Phylogenetic relationships among members of Saxifragaceae sensu lato based on rbcL sequence data. Annals of the Missouri Botanical Garden 80: 631-660.[CrossRef][ISI]
Nei M. 1987. Molecular evolutionary genetics Columbia University Press New York, New York, USA.
Ouborg N. J Piquot Y Van Groenendael J. M. 1999. Population genetics, molecular markers, and the study of dispersal in plants. Journal of Ecology 87: 551-568.[CrossRef]
Posada D Crandall K. A. 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817-818.
Remington C. 1968. Suture-zones of hybrid interaction between recently joined biotas. Evolutionary Biology 2: 321-428.
Rieseberg L. H. 1997. Hybrid origins of plant species. Annual Review of Ecology and Systematics 28: 359-389.[CrossRef][ISI]
Rieseberg L. H Wendel J. F. 1993. Introgression and its consequences in plants. In R. G. Harrison [ed.], Hybrid zones and the evolutionary process, 71–109. Oxford University Press, Oxford, UK.
Rodríguez F Oliver J. L Marín A Medina J. R. 1990. The general stochastic model of nucleotide substitution. Journal of Theoretical Biology 142: 485-501.[ISI][Medline]
Rozas J Sánchez-DelBarrio J. C Messeguer X Rozas R. 2003. DnaSP, DNA polymorphism analysis by the coalescent and other methods. Bioinformatics 19: 2496-2497.
Saitou N Nei M. 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Molecular Biology and Evolution 4: 406-425.[Abstract]
Sang T Crawford D. J Stuessy T. F. 1997. Chloroplast DNA phylogeny, reticulate evolution and biogeography of Paeonia (Paeoniaceae). American Journal of Botany 84: 1120-1136.[Abstract]
Schaal B. A Hayworth D. A Olsen K. M Rauscher J. T Smith W. A. 1998. Phylogeographic studies in plants: problems and prospects. Molecular Ecology 7: 465-474.[CrossRef]
Schneider S Roessli D Excoffier L. 2000. Arlequin ver. 2.000: a software for population genetics data analysis Genetics and Biometry Laboratory University of Geneva, Geneva, Switzerland.
Schopmeyer C. S. 1974. Seeds of woody plants in the United States USDA handbook no. 450, Washington D.C., USA.
Sewell M. M Qiu Y. L Parks C. R Chase M. W. 1993. Genetic evidence for trace paternal transmission of plastids in Liriodendron and Magnolia (Magnoliaceae). American Journal of Botany 80: 854-858.[CrossRef][ISI]
Slatkin M. 1985. Rare alleles as indicators of gene flow. Evolution 39: 53-65.[CrossRef][ISI]
Stace C. A. 1987. Hybridization and the plant species. In K. M. Urbanska [ed.], Differentiation patterns in higher plants, 115–127. Academic Press, New York, New York, USA.
Swofford D. L. 2003. PAUP*: phylogenetic analysis using parsimony (*and other methods). Version 4 Sinauer Sunderland, Massachusetts, USA.
Tajima F. 1983. Evolutionary relationships of DNA sequences in finite populations. Genetics 105: 437-460.
Thompson J. D Gibson T. J Plewniak F Jeanmougin F Higgins D. G. 1997. The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Research 24: 4876-4882.
Wagner W. H. 1969. The role and taxonomic treatment of hybrids. Bioscience 19: 785-789.[CrossRef][ISI]
Wagner W. H. 1970. Biosystematics and evolutionary noise. Taxon 19: 146-151.[CrossRef]
Weir B. S. 1996. Genetic data analysis II: methods for discrete population genetic data Sinauer Sunderland, Massachusetts, USA.
Weir B. S Cockerham C. C. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 1358-1370.[CrossRef][ISI]
Wright S. 1931. Evolution in Mendelian populations. Genetics 16: 97-159.
Xiang J. Q.-Y Crawford D. J Wolfe A. D Tang Y. -C dePamphilis C. W. 1998. Origin and biogeography of Aesculus L. (Hippocastanaceae): a molecular phylogenetic perspective. Evolution 52: 988-997.[CrossRef][ISI]
| ||||||||||||||||||||||||||||||||||||
