HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (8) Citing Articles via Google Scholar Google Scholar Articles by Kim, K.-J. Articles by Jansen, R. K. Search for Related Content PubMed Articles by Kim, K.-J. Articles by Jansen, R. K. Agricola Articles by Kim, K.-J. Articles by Jansen, R. K.

A chloroplast DNA phylogeny of lilacs (Syringa, Oleaceae): plastome groups show a strong correlation with crossing groups¹

^a Department of Biology, Yeungnam University, Keongsan, Keongbuk, Korea 712-749; ^b Department of Botany and Institute of Cell and Molecular Biology, University of Texas, Austin, Texas 78713

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Phylogenetic relationships and genomic compatibility were compared for 60 accessions of Syringa using chloroplast DNA (cpDNA) and nuclear ribosomal DNA (rDNA) markers. A total of 669 cpDNA variants, 653 of which were potentially phylogenetically informative, was detected using 22 restriction enzymes. Phylogenetic analyses reveal four strongly supported plastome groups that correspond to four genetically incompatible crossing groups. Relationships of the four plastome groups (I(II(III,IV))) correlate well with the infrageneric classification except for ser. Syringa and Pinnatifoliae. Group I, which includes subg. Ligustrina, forms a basal lineage within Syringa. Group II includes ser. Syringa and Pinnatifoliae and the two series have high compatibility and low sequence divergence. Group III consists of three well-defined species groups of ser. Pubescentes. Group IV comprises all members of ser. Villosae and has the lowest interspecific cpDNA sequence divergences. Comparison of cpDNA sequence divergence with crossability data indicates that hybrids have not been successfully generated between species with divergence greater than 0.7%. Hybrid barriers are strong among the four major plastome groups, which have sequence divergence estimates ranging from 1.096 to 1.962%. In contrast, fully fertile hybrids occur between species pairs with sequence divergence below 0.4%. Three regions of the plastome have length variants of greater than 100 bp, and these indels identify 12 different plastome types that correlate with phylogenetic trees produced from cpDNA restriction site data. Biparentally inherited nuclear rDNA and maternally inherited cpDNA length variants enable the identification of the specific parentage of several lilac hybrids.

Key Words: crossing group • genomic compatibility • lilacs • plastome group • Oleaceae • sequence divergence • Syringa

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Lilacs (Syringa) are one of the most widely cultivated ornamental trees and shrubs in temperate regions of the world. Several hundred cultivars have been developed from 22 to 28 wild species through extensive hybridization and artificial selection (McKelvey, 1928; Sax, 1945; Pringle, 1981; Fiala, 1988). Syringa has a native distribution in temperate regions of Eurasia, especially northeastern Asia and southeastern Europe (Fig. 1). Two wild species occur in eastern Europe (Green, 1972), and the majority of the species occur in northeastern Asia (Lee, 1979; Chang et al., 1992; Yamazaki, 1993). Syringa is divided into two subgenera and one of these, subg. Ligustrina, only includes two species (S. reticulata and S. pekinensis). The second subgenus, Syringa, is divided into the four series Syringa, Villosae, Pubescentes, and Pinnatifoliae (Rehder, 1945; Harborne and Green, 1980).

View larger version (29K): [in this window] [in a new window]   Fig. 1. Disjunct distribution pattern of Syringa in Eastern Europe and Asia; S. vulgaris and S. josikaea occur in southeastern Europe and the remaining 20–26 species in Asia, especially in China and Korea.

Many previous studies have demonstrated congruence among phylogenetic trees generated separately by both morphological and chloroplast (cpDNA) data (reviewed in Olmstead and Palmer, 1994). Some conflicts have been observed, especially in plant groups having a high incidence of interspecific hybridization (reviewed in Soltis et al., 1991; Rieseberg and Brunsfeld, 1992; Soltis and Kuzoff, 1995). Crossing data, however, are rarely compared to cpDNA data because both types of data are often unavailable (e.g., Doyle, Doyle, and Brown, 1990). An abundance of crossing information for Syringa provides an opportunity to evaluate the relationship between plastome and crossing groups.

In this paper, we examine phylogenetic relationships among species of lilacs using cpDNA restriction site analysis. We identify the parents of various putative interspecific hybrids using maternally inherited cpDNA and biparentally inherited nuclear ribosomal DNA (rDNA). The relationship between levels of cpDNA divergence and genetic compatibility based on the crossing data is also examined. Finally, we discuss several systematic issues in Syringa, including the origin of disjunct distribution patterns and the evolution of selected morphological and chemical characters.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Sources of plant materials for the 58 samples of Syringa and eight outgroup genera included in this study are listed in Table 1. Total DNAs were extracted from 2 to 5 g of leaf tissue pulverized in liquid nitrogen following the standard CTAB method of Doyle and Doyle (1987). For species in ser. Pubescentes, which contained large amounts of mucilage, up to 20 g of fresh leaves were ground with 200 mL of sorbitol extraction buffer (Palmer, 1986) and filtered through three layers of miracloth. Total cell organelle pellets were recovered by centrifugation (10 000 g, 10 min, 4°C), and DNAs were then extracted from the pellets using the CTAB method. All DNAs were further purified by cesium chloride/ethidium bromide gradients.

A data matrix encoding the presence or absence of each restriction site for each cpDNA was generated from the maps. Regions containing insertions or deletions (indels) greater than 100 bp were not included (see Results) due to the difficulty in determining homology of sites. For the nuclear rDNA study, a clone containing the entire Helianthus repeat (provided by M. Arnold) was used for the filter hybridizations. Nuclear rDNA data were used only for the identification of hybrids.

Eight genera of Oleaceae, Abeliophyllum, Forsythia, Chionanthus, Fraxinus, Ligustrum, Olea, Osmanthus, and Parasyringa, were included in phylogenetic analyses as possible outgroups. Forsythia and Abeliophyllum were initially selected as remote outgroups because they are in the subfamily Jasminoideae, whereas all other genera are in the subfamily Oleoideae with Syringa (Taylor, 1945; Johnson, 1957). Preliminary phylogenetic analyses using these outgroups identified both Parasyringa and Ligustrum as the sister groups to Syringa. Thus, only these two genera were used as outgroups in subsequent phylogenetic analyses.

Phylogenetic analyses were performed with Wagner parsimony using PAUP (Swofford, 1993; version 3.1.1). The amount of phylogenetic signal was evaluated using the G₁ value (Hillis, 1991) and the cladistic permutation tail probability (PTP) test (Faith and Cranston, 1991; see below). Heuristic searches employed the tree bisection reconnection (TBR) branch swapping to find the most parsimonious (MP) trees for the data set of all 60 taxa. To identify multiple islands of equally parsimonious trees (Maddison, 1991), 1000 random entries were performed. Bootstrap analysis (Felsenstein, 1985) included 1000 replicates using TBR branch swapping without MULPARS and either accelerated (ACCTRAN) or delayed character transformations (DELTRAN).

The data set was reduced to 24 taxa, including 22 ingroup and two outgroup species for more rigorous phylogenetic analyses. The branch-and-bound search option (Hendy and Penny, 1982) was employed for the reduced data set. Support for each clade in the 24-taxa tree was evaluated using three methods, bootstrapping (Felsenstein, 1985; Sanderson, 1989), decay analysis (Bremer, 1988, 1994; Hillis and Dixon, 1989), and the topology-dependent cladistic permutation tail probability (T-PTP) test (Faith, 1991). Bootstrap analyses were repeated 1000 times using TBR without MULPARS and either ACCTRAN or DELTRAN.

For the PTP (Faith and Cranston, 1991) and T-PTP tests (Faith, 1991) all autapomorphic characters were removed from the data matrix and the remaining characters were permuted among taxa. The 99 randomized data sets were generated by a Macintosh version of randomization software (written by J. Huelsenbeck). Outgroup taxa were added to the data matrix after randomization according to the suggestion of Faith and Cranston (1991). An original and 99 randomized data sets were analyzed as a single large batch file of PAUP. Branch-and-bound searches were used for PTP tests. As an indicator of significance of cladistic structure, the PTP value (Faith and Cranston, 1991) was calculated by comparing tree lengths from the original and randomized data sets.

All clades in the reduced tree were subjected to the T-PTP test. The monophyly and nonmonophyly of each clade were evaluated for the original and 99 randomized data sets. A total of 42 000 PAUP analyses (100 data sets x 21 clades x 2 for different character transformations x 10 random entries) were performed using the TBR option. To calculate T-PTP values, the tree length difference between the monophyly and nonmonophyly tests of each clade (calculated as the minimum length under nonmonophyly minus the minimum length under monophyly) was first determined for the original data set. Following this step, length differences between the monophyly and nonmonophyly tests of each clade were calculated for the 99 randomized data sets. Then, length differences in each clade for the original data set were compared to those of the 99 randomized data sets. The proportion of length difference values from 99 randomized data sets equal or larger than the difference value from the original data was referred to as the T-PTP value (Faith, 1991). A T-PTP value equal to or greater than 0.05 was considered as a clade that does not have significant support for its monophyly.

Sequence divergence values between species were calculated using Eqs. 9 and 10 of Nei and Li (1979) for all pairwise comparisons among the 22 ingroup taxa. Different r values were employed for the 5- and 6-bp (base pair) recognition enzymes, but r values were not corrected for enzymes that recognize multiple sequences. As a result, sequence divergence values may be slightly underestimated (Nei, 1987).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Chloroplast DNA restriction site variation
The 22 enzymes identified an average of 1226 restriction sites in the chloroplast genome of each species, excluding the three regions with substantial length variation (see next section). This represents a sampling of 5.7% of the genome. Nine hundred and fifty invariant sites were detected among all taxa of Syringa. The number of variable sites among accessions of Syringa and the two outgroup genera was 669, 653 of which were potentially phylogenetically informative (a copy of the data matrix is deposited in TEX/LL and is available from K.-J. Kim).

The 669 variable restriction sites were used as characters in phylogenetic analyses. Wagner analyses of all 60 taxa produced two equally parsimonious trees with a length of 919 (including autapomorphies), a consistency index (CI) of 0.723 (excluding autapomorphies), and a retention index (RI) of 0.953. The two trees differed only in the relative positions of S. meyeri, S. microphylla and a putative hybrid between these species. As a result, the strict consensus tree showed a trichotomy at the node involving these three taxa (Fig. 2). The cpDNA tree revealed four plastome clades that correspond to four crossing groups (see bold lines in Fig. 2). Group I was basal in Syringa and included the two tree species, S. pekinensis and S. reticulata, and their varieties of subg. Ligustrina. The two species of this subgenus showed substantially divergent plastome types from the other three groups, differing by at least 71 restriction site changes (Fig. 2). Plastome groups II–IV were monophyletic, supported by 26 synapomorphies and a 94% bootstrap value. These three groups corresponded to subg. Syringa. Group II included the monotypic ser. Pinnatifoliae and all species of ser. Syringa. Groups III and IV were represented by ser. Pubescentes and Villosae, respectively.

View larger version (51K): [in this window] [in a new window]   Fig. 2. Strict consensus tree of two most parsimonious trees based on chloroplast DNA restriction site analysis. The tree has 919 steps, a CI of 0.723 (without autapomorphies), and a RI of 0.953. Bold lines circumscribe four plastome groups (I, II, III, and IV), which correspond to four crossing groups. Numbers of supporting characters in ACCTRAN are given above each node and bootstrap percentages are given below. Multiple accessions for each taxon are numbered in the order given in Table 1 . Arrowheads indicate 12 length variants in the 5' region of accD and the noncoding regions between trnN and ndhF and atpH and atpI . Three deletions (-1200 bp, -600 bp, and -300 bp) in the 5' region of accD are mapped.

View larger version (38K): [in this window] [in a new window]   Fig. 3. The single most parsimonious tree from a 24-taxa analysis based on the chloroplast DNA restriction site data. The tree was generated by the branch-and-bound option and has 832 steps, a CI of 0.722 (without autapomorphies), and a RI of 0.897. Arabic numbers (1–21) above each node indicate the clade numbers in Table 2 . Clade support values including the number of supporting characters, bootstrap percentage, decay index, and T-PTP, are given in Table 2 . Letters A-I on nodes indicate the direction of change of seven morphological characters and one chemical compound. Black and shaded bars indicate nonhomoplastic and homoplastic character changes, respectively. Character transitions are: A = shrub tree, B = short long corolla tubes, C = inserted exserted anthers (reversal), D = simple compound leaves (reversal) or simple compound or laciniate leaves (parallelism), E = presence absence of hairs with mucilage, F = lateral terminal position of inflorescences (reversal), G = erect pendulous habit of panicles, and H = presence absence of flavone glycosides (parallelism).

There were 21 clades in the 24-taxa trees (Fig. 3). The degree of clade support was evaluated by the number of supporting characters both with ACCTRAN and DELTRAN character-state optimization, the numbers of nonhomoplastic characters, decay index, bootstrap percentages, and the T-PTP values (Table 2).

Ten of 21 clades had 100% bootstrap values (Table 2), including the four plastome groups and several clades within each group. Clades supported by one or two characters had low bootstrap values (clades 18 and 21). In addition, some clades (5, 9, and 14) with more than nine character changes also had relatively low (less than 80%) bootstrap values. In contrast, clade 20, which was supported by only three homoplastic characters, had a 96% bootstrap value.

There were 4, 8, 21, 49, 81, 133, 228, 318, 458, and 675 trees at 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10 step(s) longer than the MP tree, respectively. Six clades (5, 9, 14, 18, 20, and 21) collapsed in trees with fewer than four added steps. Ten clades, including all those leading to the four major plastome groups, had decay indices >10 (Table 2).

Monophyly of all clades except node 21 was supported significantly by the T-PTP tests (Table 2). The monophyly of clade 21 was rejected nine times from 99 randomized data sets (T-PTP value = 0.10). This was the only node that was not supported at a 0.05 probability level (95% confidence).

Length variation
Several regions of the Syringa chloroplast genome showed length variation, but most indels were <100 bp. Only three regions had indels >100 bp (Table 3), and these were omitted from phylogenetic analyses because of the uncertainty of homology of restriction sites. However, the indels provided independent phylogenetic information, especially for identification of hybrids. Length variation in the accD region illustrated group-specific deletion patterns (Fig. 4). Plastome groups I and II (Fig. 2) shared a 1400-bp fragment with the outgroup genera. Members of plastome group III shared a 1200-bp deletion supporting the monophyly of ser. Pubescentes. Plastome group IV (ser. Villosae) had taxa with fragment patterns of 800 bp and 500 bp (Fig. 2). The most parsimonious interpretation of these indels suggests that two independent deletions (600 and 300 bp) occurred in ser. Villosae. A 600-bp deletion originated prior to the diversification of the series and a subsequent 300-bp deletion occurred in the clade leading to the group of species including S. yunnanensis–S. tomentella. The two fragment patterns in ser. Villosae (plastome group IV) were particularly useful for identifying the maternal parentage of hybrids, which has often been difficult for lilac breeders.

View larger version (51K): [in this window] [in a new window]   Fig. 4. Length variation in the 5' region of accD gene from 33 representative Syringa taxa. BamH I digested DNAs were hybridized with P-labeled tobacco probe 19 (Olmstead and Palmer, 1992). Location and size of length variants are given in the restriction map at bottom. Numbers on right indicate the approximate fragment sizes in kilobases. Taxon names for each lane are: 1 = S. pubescens (plastome group III); 2 = S. meyeri (III); 3 = S. patula (III); 4 = S. pekinensis (I); 5 = S. pekinensis var. pendula (I); 6 = S. reticulata (I); 7 = S. reticulata var. manshurica (I); 8 = S. vulgaris (II); 9 = S. oblata (II); 10 = S. oblata var. dilatata (II); 11 = S. diversifolia (II); 12 = S. laciniata (II); 13 = S. pinnatifolia (II); 14 = S. emodii (IV); 15 = S. emodii (IV); 16 = S. josikaea (IV); 17. S . x josiflexa (IV); 18 = S. reflexa (IV); 19 = S. sweginzowii (IV); 20 = S. tigerstedtii (IV); 21 = S. tomentella (IV); 22 = S. villosa (IV); 23 = S. wolfii (IV); 24 = S. yunnanensis (IV); 25 = S . x henryi x villosa (IV); 26 = S . x henryi (IV); 27 = S . x nanceana (IV); 28 = S . x prestoniae `Isabella' (IV); 29 = S . x prestoniae `Olivia' (IV); 30 = S . x swegiflexa (IV); 31 = S . x swegiflexa (IV); 32 = S. emodii x sweginzowii (IV); and 33 = S . x sweginbretta (IV).

The third indel region was between the atpH and atpI genes in the large single-copy region. Eight fragment patterns were observed that can be grouped into the four plastome groups. Plastome group I had three indels in this region that differed by 200 bp. Groups II and III showed group-specific length variants and group IV revealed three distinct patterns of indels (Table 3).

The combination of three variants from three plastome regions identified a total of 12 indels (Table 3) that correlated with the groups in phylogenetic trees generated from cpDNA restriction site data (see arrowheads in Fig. 2). Three indels in group I are taxon specific. No length variation was detected in plastome group II. Length variants defined two and six groups of species in groups III and IV, respectively (Fig. 2).

Ribosomal DNA variation
Only ten of the 22 enzymes (AseI, BamHI, BclI, BstNI, DraI, EcoRI, EcoRV, NciI, SspI, and XbaI) had interpretable fragment patterns. The length of the rDNA repeat in Syringa ranged from 9.4 to 10.0 kb. Our survey included 20 accessions of putative hybrids, including two interseries hybrids. Additive patterns of nuclear rDNA were observed for several hybrids. However, one interseries hybrid (S. laciniata x S. reticulata) and four putative hybrids within series Villosae did not exhibit additivity of rDNA fragments (Table 4). Three different rDNA inheritance patterns were observed among the ten putative hybrids examined in ser. Villosae (Fig. 5). Two accessions (S. emodii x sweginzowii and S. x sweginbretta) showed unique rDNA types compared to their parents. Observed patterns of cpDNA and rDNA are summarized in Table 4.

View larger version (88K): [in this window] [in a new window]   Fig. 5. Nuclear rDNA fragment patterns of Syringa series Villosae and two other species. Ase I, Ssp I, and Nci I digested DNAs were hybridized with a P-labeled Helianthus rDNA probe. Numbers on right indicate the approximate fragment size in kilobases. Taxon names for each lane are: 1 = S. pinnatifolia ; 2 = S. emodii ; 3 = S. emodii ; 4 = S. josikaea ; 5 = S . x josiflexa ; 6 = S. reflexa ; 7 = S. sweginzowii ; 8 = S. tigerstedtii ; 9 = S. tomentella ; 10 = S. villosa ; 11 = S. wolfii ; 12 = S. yunnanensis ; 13 = S . x henryi x villosa ; 14 = S . x henryi ; 15 = S . x nanceana ; 16 = S . x prestoniae `Isabella'; 17 = S . x prestoniae `Olivia'; 18 = S . x swegiflexa ; 19 = S . x swegiflexa ; 20 = S. emodii x sweginzowii ; and 21 = S . x sweginbretta ; and 22 = S. pinetorum .

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Morphological species boundaries are controversial within subg. Ligustrina. Green and Chang (1995) recognized a single species (S. reticulata) with three subspecies. All three subspecies have been previously treated as species or varieties (Lingelsheim, 1920; McKelvey, 1928; Rehder, 1977). The taxa have a disjunct distribution in three regions of northeast Asia, northern Japan (subsp. amurensis), Korea and Manchuria (subsp. reticulata), and northern China (subsp. pekinensis). In the cpDNA tree, 46 and 36 characters support the monophyly of S. pekinensis and S. reticulata, respectively (Fig. 2). In addition, two varieties of S. reticulata (var. reticulata and var. manshurica) are also distinguished by 23 restriction site changes. It is notable that there are no successful hybrids between S. reticulata and S. pekinensis even though this cross has often been attempted (Pringle, 1981). The large number of restriction site differences among the three subspecies (sensu Green and Chang, 1995), especially in comparison with Syringa species in other branches of the cpDNA tree, supports their recognition as distinct species.

Subgenus Syringa is divided into four series, Pinnatifoliae, Syringa (= Vulgares), Pubescentes, and Villosae (Rehder, 1945). Series Pinnatifoliae has been separated from ser. Syringa because of its distinctive pinnately compound leaves (Rehder, 1945), although there is considerable genetic compatibility between these two series (Sax, 1945; Pringle, 1981). Only a single species, S. pinnatifolia, is included in the ser. Pinnatifoliae. Our cpDNA data indicate that this species is sister to ser. Syringa, and together they form a monophyletic group (plastome group II) supported by 35 characters and 100% bootstrap value (Fig. 2). The monophyly of ser. Syringa is supported by a 53% bootstrap value and S. pinnatifolia is nested within ser. Syringa in many of the bootstrap replicates. In addition, successful hybridization between the two series indicates a high degree of genetic compatibility (Sax, 1945; Pringle, 1981). Thus, the cpDNA data support the merger of the ser. Pinnatifoliae and Syringa.

Two strongly supported clades comprise ser. Syringa (Fig. 2). One clade includes the laciniate-leaved species and their hybrids, including S. laciniata (S. protolaciniata sensu Green, 1989) and S. x chinensis. The laciniate-leaved lilacs have been assigned various names, such as S. laciniata, S. afghanica, and S. persica var. laciniata (Lingelsheim, 1920; McKelvey, 1928; Sax, 1945). Green (1989) identified fertile and sterile lineages of laciniate-leaved lilacs and applied a new name, S. protolaciniata, for the former and S. x laciniata for the latter. He also suggested that S. x laciniata was probably derived from a hybrid between S. protolaciniata and S. vulgaris. Several studies also indicated that the two entire-leaved hybrids, S. x chinensis and S. x persica, are derived from a cross between S. laciniata (S. protolaciniata sensu Green, 1989) and S. vulgaris (McKelvey, 1928; Sax, 1945; Lemoine, 1990; Marsolais, Pringle, and White, 1993). However, it is still unclear how many distinct variants of S. laciniata sensu lato (s.l.) exist in both natural habitats and cultivation, and how various leaf forms of S. chinensis, S. persica, and S. x laciniata are derived from the same or similar parental species. The second clade in ser. Syringa includes the two most widely cultivated species of lilacs, S. vulgaris and S. oblata. Several hybrids between these two species form a monophyletic group either with S. vulgaris or S. oblata, indicating strong genetic compatibility in both directions.

Series Pubescentes (plastome group III) is the most strongly supported clade with 85 characters and a bootstrap value of 100% (Figs. 2, 3). The cpDNA tree identifies three well-supported monophyletic groups in this plastome group: S. pubescens; S. julianae–S. microphylla; and S. potaninii–S. patula. The first group includes a single species, S. pubescens, and is sister to the other two groups. In the second group, S. julianae is sister to the S. meyeri–microphylla complex. The third species group includes S. potaninii and S. patula and has virtually no cpDNA variation. Taxonomic circumscriptions in ser. Pubescentes have been controversial since McKelvey's (1928) recognition of ten species (Rehder, 1928; Chang et al., 1992; Green and Chang, 1995). Green and Chang (1995) recently developed a broad definition of S. pubescens, which included S. patula, S. microphylla, S. julianae, and S. potaninii. These same workers recognized S. mairei, S. pinetorum, S. wardii, and S. meyeri as distinct species. Several differences are apparent between the cpDNA tree (Fig. 2) and Green and Chang's (1995) taxonomic treatment. For example, the merger of S. microphylla with S. pubescens and the recognition of S. meyeri as a distinct species are not congruent with the cpDNA tree. These incongruences cannot be resolved without further study.

Series Villosae of subg. Syringa consists of 7–10 species. Our cpDNA study sampled 26 accessions of the series, including all ten native species and ten interspecific hybrids. Monophyly of the series is supported by 49 characters and a 100% bootstrap value (Fig. 2). The cpDNA tree identifies three well-supported monophyletic groups within the series. The first only includes S. emodii, which is the sister species to the two other groups. The second group includes three wild species, S. villosa, S. wolfii, and S. josikaea, and several interspecific hybrids. The two varieties of S. wolfii form a paraphyletic group. The third group includes six species with the following relationships: (S. yunnanensis (S. komarowii, S. reflexa)(S. sweginzowii (S. tomentella, S. tigerstedtii))). However, support for some of the nodes is weak. The monophyly of S. komarowii and S. reflexa is strongly supported, which is consistent with the recent merger of these two species (Green and Chang, 1995).

Evolution of morphological and chemical characters
Most species of Syringa are defined by quantitative characters, such as shape and size of leaves, flowers, and fruits. Only seven discrete morphological characters have been used in classifications of the genus. Thus, our discussion of character evolution will focus only on these seven morphological characters and one class of chemical compounds. Overall, four of the eight characters exhibit homoplasy when mapped onto the cpDNA tree (Fig. 3).

Both trees and shrubs occur in Syringa. Members of subg. Syringa have a shrubby habit both in cultivation and in their native habitat. In contrast, species in subg. Ligustrina are trees. For example, individuals of S. reticulata with unbranched stems >20 m are observed frequently in temperate deciduous forests of Korea and northern China (K.-J. Kim, personal observation). However, trees are uncommon in cultivation. The cpDNA phylogeny suggests that the tree habit is a synapomorphy of subg. Ligustrina (character A in Fig. 3).

Variation in length of the corolla tube and filament has produced two distinct floral forms in Syringa. The first type, which has exserted anthers and short corolla tubes, occurs only in subg. Ligustrina and the sister genus Ligustrum. In contrast, all members of subg. Syringa, except S. emodii, have long corolla tubes with short filaments with subsessile anthers. In S. emodii, the corolla tube is still relatively long, but the anthers are slightly exserted. The cpDNA tree suggests that the long corolla tube and inserted anthers in subg. Syringa are derived and that the transition occurred prior to the diversification of the subgenus (characters B and C in Fig. 3). Subsequently, there was a reversal in the anther character without modification of the corolla tube in S. emodii.

Only S. pinnatifolia has pinnately compound leaves and one or two species have laciniate leaves (Green, 1989). According to the cpDNA tree the compound leaf of S. pinnatifolia is derived from an ancestor with simple leaves before the diversification of plastome group II. A subsequent reversal led to the simple-leaved condition in ser. Syringa through the intermediate laciniate-leaved condition (character D in Fig. 3). A second equally parsimonious interpretation from the cpDNA tree is that the compound leaves in S. pinnatifolia and laciniate leaves in S. laciniata are derived independently from simple leaved ancestors.

Trichomes are rare on leaf surfaces of lilacs except in ser. Pubescentes. The cpDNA tree suggests that this feature is an apomorphic condition that characterizes this group (character E in Fig. 3).

Inflorescences of many lilac species develop from two lateral buds rather than a terminal bud. As a result, the panicles appear as a Y-shaped cluster of flowers at the tips of branches. All species of Syringa except ser. Villosae share the same developmental pattern, although the patterns are less obvious during the late flowering stages because the growing terminal bud may change into a vegetative branch. The panicles of ser. Villosae usually develop from a single terminal bud, which also generates several basal leaves. Two lateral buds usually develop into vegetative branches in ser. Villosae. This type of developmental pattern occurs in nine of the ten species of ser. Villosae. A modification of this pattern occurs in S. sweginzowii, where both terminal and lateral buds develop into inflorescences. The cpDNA tree suggests that a terminal inflorescence present in plastome group IV is a derived condition (Figs. 2, 3). Thus, the presence of two lateral inflorescences in S. sweginzowii of ser. Villosae represents a reversal (character F in Fig. 3).

Most species of Syringa have erect or semi-erect panicles. Two members of ser. Villosae, S. reflexa and S. komarowii, have distinctive pendulous panicles. The cpDNA tree suggests that the pendulous panicle is derived from an erect panicle (character G in Fig. 3).

Flavonol glycosides are widespread in Syringa, whereas flavone glycosides are more narrowly restricted (Harborne and Green, 1980). The distribution of these chemical data on the cpDNA tree indicates that flavone glycosides are restricted to ser. Ligustrina, Pinnatifoliae, and Syringa and were lost prior to the diversification of ser. Pubescentes and Villosae (character H in Fig. 3). In addition, a parallel loss of flavone glycosides occurred in the branch leading to S. vulgaris.

Biogeography
Syringa is restricted to temperate regions of southeastern Europe and Asia (Fig. 1). Only two species, S. vulgaris and S. josikaea, are distributed in southeastern Europe and are disjunct from the majority of their northeast Asian counterparts. The two European species are positioned in distant lineages in the cpDNA tree (Figs. 2, 3). Syringa vulgaris forms a well-supported monophyletic group with S. oblata of northern China and Korea, whereas S. josikaea is positioned between two varieties of S. wolfii from the Korean peninsula. These disparate placements are not surprising because the European species have always been classified into different series (Rehder, 1928, 1945) and they show strong genetic incompatibility (Pringle, 1981). Constraining the monophyly of the European species in the cpDNA tree requires 162 more steps than the most parsimonious tree. Therefore, the cpDNA data strongly support independent origins of the two European species.

The disjunct distribution of the European species from their sister species in northeastern Asia may have originated by independent long-distance dispersal events. Alternatively, the ancestors of the two European taxa may have been widely distributed from northeastern Asia to eastern Europe and subsequent extinction in central Asia generated the current disjunct distribution. The second hypothesis is supported by fossil species (leaf impressions) of Syringa from the Tertiary to the middle Pleistocene of the Quaternary in Hungary (Andreanszky, 1968; Skoflek, 1968). The fossils include S. palaeojosikaea, which resembles the extant species S. josikaea. If we use the average cpDNA sequence divergence rate of 0.1% per million years following Parks and Wendel (1990) and a sequence divergence of 1.204%, the two European species would have diverged from each other ~12 million years ago. This corresponds to the early Pliocene of the Tertiary. The estimated divergence time corresponds well with the fossil leaf remains reported from Tertiary in Hungary (Andreanszky, 1968). If this estimate is accurate, the current disjunct distribution of closely related species in southeastern Europe and northeastern Asia probably reflects the glacial contraction and interglacial expansion of the distribution ranges. A similar disjunct distribution pattern in northeastern Asia and southeastern Europe has been identified in related species pairs of Forsythia, also in the Oleaceae (K.-J. Kim, unpublished data). In addition, the current distributions of the two European Syringa species do not overlap. Syringa vulgaris is distributed widely in southern parts of eastern Europe from north-central Romania to central Albania and northeastern Greece, whereas S. josikaea is restricted to the mountains of Transylvania and the Ukrainian Carpathians (Green, 1972).

The geographic distributions of the most closely related Asian species also do not overlap. Three taxa in plastome group I have allopatric distributions in northern China, Korea, and Japan (Green and Chang, 1995). Most species in plastome group II are also allopatric and only the S. oblata and S. laciniata (S. protolaciniata sensu Green, 1989) species pair shows sympatry in Gansu and Ninxi provinces of China. However, artificial hybrids between these taxa are fully sterile (Pringle, 1981) and there are no reported natural hybrids.

Members of plastome group III are concentrated primarily in northern China. Syringa microphylla and S. pubescens are distributed in largely overlapping areas of northern China, whereas the other species (S. patula, S. potaninii, S. meyeri, S. wardii, etc.) have allopatric distributions. Therefore, natural hybridization is possible only between two of the species. However, there are no reported natural or artificial hybrids between S. microphylla and S. pubescens.

The ten species of group IV are concentrated primarily in southern China, although some species occur northward to the Korean peninsula and westward to Afghanistan. The distributions of the species of ser. Villosae do not overlap, even in the Yunan and Sichuan provinces of China where six species occur in different habitats. Thus, geographical isolation is the primary mechanism keeping these fully genetically compatible species apart.

Identification of hybrids in Syringa and the use of DNA data as an indicator for future breeding experiments
Several hundred lilac cultivars have been generated by extensive hybridization, cultivation of chance hybrid seedlings in nurseries, and artificial selection. It was not our intention to verify all hybrids in cultivation, but we included several putative hybrids in the study. Although the origin of these is controversial among lilac breeders, many are cultivated widely in gardens and are utilized in crossing experiments. Some of the uncertainty regarding the origin of hybrids is due, at least in part, to the fact that the uniform chromosome numbers (N = 23) and small chromosome sizes (Sax, 1930, 1947; Taylor, 1945) has rendered these data essentially useless for identification of the parents.

Most hybrids of Syringa show an additive pattern of the nuclear rDNA types of their putative parents. Different rDNA types in hybrids can be easily maintained because of the predominance of vegetative propagation. Homogenization of rDNA units via gene conversion or unequal crossing over (Dover, 1982; Hillis et al., 1991) would not operate in the clonal lines of hybrids. Thus, rDNA data can provide conclusive evidence for the parentage of hybrids if parental lilac species have different rDNA types. This may not be the case in fully fertile hybrids because of concerted evolution (Dover, 1982; Hillis et al., 1991; Wendel, Schnabel, and Seelanan, 1995).

Comparison of data from biparentally inherited rDNA and maternally inherited cpDNA can allow identification of the specific maternal and paternal parents of hybrids. Length variants in cpDNA are particularly useful for identifying the maternal parent of hybrids if the parents have different size fragments. For example, the maternal genome donor of hybrids between two groups of ser. Villosae (S. emodii–villosa–wolfii–josikaea species group and S. yunnanensis–komarowii–reflexa–sweginzowii–tomentella–tigerstedtii species group) can be identified using indels in the accD region (Fig. 4).

There has been considerable controversy regarding the correct identity of hybrids between subg. Ligustrum and Syringa or between different series (Sax, 1945; Pringle, 1981). One putative intersubgeneric hybrid, S. reticulata x S. laciniata, was included in our DNA studies. Syringa reticulata and S. laciniata show numerous cpDNA and nuclear rDNA restriction site differences. However, both cpDNA and rDNA fragment patterns (Table 4) of the putative hybrid were identical to S. reticulata var. manshurica. Thus, our data suggest that this putative hybrid is actually S. reticulata, which is consistent with previous suggestions that intersubgeneric hybrids do not exist in Syringa (Rehder, 1945; Pringle, 1981).

We also investigated an accession of S. x diversifolia (= S. pinnatifolia x S. oblata), which is widely recognized as a hybrid between ser. Pinnatifoliae and Syringa. The hybrid has 3–5 leaflets that are intermediate between the pinnately compound leaves of S. pinnatifolia and the simple leaves of S. oblata. Syringa x diversifolia is a vigorous F₁ hybrid propagated by cuttings because of F₁ sterility. The hybrid was produced at the Arnold Arboretum in 1929 (Anderson and Rehder, 1935) from crosses between S. pinnatifolia and S. oblata. The combined nuclear rDNA and cpDNA data indicate that S. oblata was the pollen donor and S. pinnatifolia was the maternal parent. Our cpDNA tree places the two series Pinnatifoliae and Syringa into a single plastome group, suggesting that these should probably be combined into one. Following this taxonomic merger of the series, S. x diversifolia would be more accurately described as an intraseries hybrid.

DNA data provide interesting insights into the origin of S. x chinensis. Nuclear rDNA patterns from S. x chinensis are additive with respect to S. laciniata and S. vulgaris, both in terms of length and restriction sites. However, cpDNA of S. x chinensis differ from both S. laciniata and S. vulgaris, although its cpDNA lineage was closer to S. laciniata than S. vulgaris (Fig. 2). In addition, the uniform cpDNA and rDNA patterns of four different accessions of S. x chinensis indicate that this cultivar is probably derived from similar parents by unidirectional hybridization. Our sampling includes only a portion of the genetic diversity in this complex. However, the results suggest that the currently available accessions of S. laciniata (S. protolaciniata sensu Green, 1989) were not parents of S. x chinensis.

Similar unidirectional hybridization was observed in S. x swegiflexa (= S. sweginzowii x S. reflexa), S. x prestonii (= S. reflexa x S. villosa), and the hybrid complex of S. x henryi (= S. josikaea x S. villosa). All three accessions of S. x swegiflexa have the maternal genome of S. reflexa rather than S. sweginzowii. Both accessions of S. x prestonii have the maternal genome of S. villosa rather than S. reflexa and two of three accessions of the S. x henryi complex have the S. villosa cpDNA pattern (Table 4). In contrast, S. x hyacinthiflora (= S. oblata x S. vulgaris) showed a bidirectional hybridization pattern. The two accessions of S. x hyacinthiflora had an additive nuclear rDNA pattern of S. oblata and S. vulgaris, whereas the cpDNA types matched either S. vulgaris or S. oblata.

In some instances, putative hybrids do not show combined rDNA patterns. One accession of the S. x henryi complex and S. x swegiflexa has the rDNA pattern of a single parent. The rDNA data suggest that the accessions may not be hybrids or that concerted evolution resulted in fixation of one parental type.

Clade support
Clade support indices are not always well correlated. For example, clade 20, which has no nonhomoplastic character support and a low decay index (2), has a relatively high bootstrap value (96%). Clade 9 has a relatively low bootstrap value (80%) and low decay index (3) but a high T-PTP value (<0.01). The monophyly of five clades (5, 9, 14, 18, and 21) is rejected at the 95% bootstrap level, however, the T-PTP test only rejects the monophyly of one (21) of these at a significance level of >0.05, which is the cut-off level suggested by Faith and Cranston (1991). The T-PTP test rejects the monophyly of only five clades at the highest level of confidence (0.01). Thus, the T-PTP test probably overestimates confidence of monophyletic groups compared to other clade support values. We do not know whether the comparisons in Syringa will apply to other groups. More extensive examination of clade support indices from different data sets will reveal general correlations among the numbers of supporting characters, bootstrap percentages, decay indices, and the T-PTP values.

Correlation between plastome groups and crossing groups
We categorized Syringa crossing data into four classes: no successful hybrids; F₁ does not reach flowering stage; F₁ is vigorous but sterile; and hybrids are fully fertile. There is a strong correlation between the four plastome groups and the four crossing groups (Fig. 2). Chloroplast DNA divergence among and within the four plastome groups is summarized in Table 4. These data indicate clearly that there are no reported hybrids between taxa with more than 0.7% cpDNA sequence divergence. Hybridization between different plastome groups is unconfirmed and also highly unlikely given that sequence divergences among the four plastome groups range from 1.096 to 1.962%. Restricted hybridization among series is also demonstrated with the absence of any confirmed interseries hybrids of lilacs except between ser. Pinnatifolia and Syringa, despite numerous crossing attempts by lilac breeders (Pringle, 1981). Our cpDNA tree (Fig. 2) indicates that ser. Pinnatifoliae and Syringa are in the same plastome group and sequence divergence between the two series ranges only from 0.466 to 0.606%.

Difficulties have been reported in intragroup crosses involving some species in plastome groups II and III (Pringle, 1981). These cases correspond to the second and third classes of crossing categories, and their cpDNA sequence divergence ranges between 0.4 and 0.7%. The fourth crossing class, fully fertile hybrids, is most commonly observed within plastome group IV, which shows the lowest cpDNA sequence divergence (below 0.500%). The boundary between fully fertile hybrids and F₁ sterile hybrids overlaps slightly depending on the combination of species. Thus, it appears that in Syringa a cpDNA sequence divergence between 0.4 and 0.5% is the cut-off point for generating fully fertile hybrids. Similar levels of correlation between cpDNA sequence divergence and genomic incompatibility were also identified in Glycine (Singh and Hymowitz, 1985a, b; Doyle, Doyle, and Brown, 1990) and Gossypium (Wendel, 1989; Wendel and Albert, 1992).

Chloroplast DNA divergence may be used as a predictor of hybridization success in future lilac breeding experiments. Interplastome group or interseries hybridizations will probably be unsuccessful except between S. pinnatifolia and species of ser. Syringa. Within plastome groups II and III, the majority of hybrid combinations will result in sterile hybrids even if they are vigorous. Sterile hybrids may have horticultural value and may be propagated by vegetative reproduction. In contrast, most interspecific hybridizations within group IV (ser. Villosae) will result in fully fertile hybrids. However, some degree of sterility will probably be encountered in hybridizations involving S. emodii, S. wolfii, and S. josikaea.

	FOOTNOTES

 
¹ The authors thank Jack Alexander and Robert Nicolson for help in the collection of living plant materials; John Huelsenbeck for assistance in data analysis; Bob Cook of the Arnold Arboretum for financial support; and John Bain, Harvey Ballard, Todd Barkman, John Clement, Carolyn Ferguson, Les Goertzen, and Doug Goldman for critical reading of the manuscript. This research was supported by a Putnam Fellowship from the Arnold Arboretum and grants from the Korea Research Foundation to KJK (DRF 01-D-0659) and National Science Foundation to RKJ (DEB-9318278).

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Anderson, E., and A. Rehder.1935.New hybrids from the Arnold Arboretum. Journal of the Arnold Arboretum 16: 358–363.

Andreanszky, G. 1968.Reste d'un lilas du sarmatien Hongrois. Acta Botanica Academiae Scientiarum Hungaricae 14: 133–145.

Bremer, K.1988.The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 795–803. [CrossRef][ISI]

———.1994.Branch support and tree stability. Cladistics 10: 295–304. [CrossRef][ISI]

Chang, M., B. Miao, R. Lu, and L. Qui.1992.Flora republicae popularis Sinicae 61: Oleaceae and Loganiaceae. Science Press, Peking.

Dover, G. A.1982.Molecular drive, a cohesive mode of species evolution. Nature 299: 111–117. [CrossRef][Medline]

Doyle, J. J., and J. L. Doyle.1987.A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11–15.

———, ———, and A. H. D. Brown.1990.A chloroplast-DNA phylogeny of the wild perennial relatives of soybean (Glycine subgenus Glycine): congruence with morphological and crossing groups. Evolution 44: 371–389. [CrossRef][ISI]

Faith, D. P.1991.Cladistic permutation tests for monophyly and nonmonophyly. Systematic Zoology 40: 366–375. [CrossRef]

———, and P. S. Cranston.1991.Could a cladogram of this sort have arisen by chance alone? On permutation tests for cladistic structure. Cladistics 7: 1–28.

Felsenstein, J.1985.Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791. [CrossRef][ISI]

Fiala, J.1988.Lilacs. Timber Press, Portland, OR.

Green, P. S.1972.Syringa. In T. G. Tutin, V. H. Heywood, N. A. Burges, D. M. Moore, D. H. Valentine, S. M. Walters, and D. A. Webb [eds.], Flora Europaea 3, 54. Cambridge University Press, Cambridge.

———.1989.The laciniated-leaved lilacs. Kew Magazine 6: 116–124.

———, and M. C. Chang.1995.Some taxonomic changes in Syringa L. (Oleaceae), including a revision of series Pubescentes. Novon 5: 329–333.

Harborne, J. B., and P. S. Green.1980.A chemotaxonomic survey of flavonoids in leaves of the Oleaceae. Botanical Journal of the Linnean Society 81: 155–167.

Hendy, M. D., and D. Penny.1982.Branch and bound algorithms to determine minimal evolutionary trees. Mathematical Bioscience 59: 227–290.

Hillis, D. M.1991.Discriminating between phylogenetic signal and random noise in DNA sequences. In M. K. Miyamoto and J. Cracraft [eds.], Phylogenetic analysis of DNA sequences, 278–294. Oxford University Press, New York, NY.

———, and M. T. Dixon.1989.Vertebrate phylogeny: evidence from 28S ribosomal DNA sequences. In B. Fernholm, K. Bremer, and H. Jornvall [eds.], The hierarchy of life, 355–367. Elsevier Science Publishers, Amsterdam.

———, and J. P. Huelsenbeck.1992.Signal, noise, and reliability in molecular phylogenetic analyses. Journal of Heredity 83: 189–195. [Abstract/Free Full Text]

———, C. Moritz, C. A. Porter, and R. J. Baker.1991.Evidence for biased gene conversion in concerted evolution of ribosomal DNA. Science 251: 308–310. [Abstract/Free Full Text]

Johnson, L. A. S.1957.Review of the family Oleaceae. Contributions from the New South Wales National Herbarium 2: 395–418.

Lee, C. B.1979.Illustrated flora of Korea. Hyangmoonsa, Seoul.

Lemoine, E.1990.Hybrids between the common lilac and the laciniated Persian lilac. Journal of the Royal Horticultural Society 24: 299–311.

Lingelsheim, A.1920.Oleaceae-Oleoideae-Syringeae. In A. Engler, [ed.], Das Pflanzenreich, 106: 5–125. Verlag von Wilhelm Engelmann, Leipzig.

Maddison, D. R.1991.The discovery and importance of multiple islands of most-parsimonious trees. Systematic Zoology 40: 315–328. [CrossRef]

Marsolais, J. V., J. S. Pringle, and B. N. White.1993.Assessment of random amplified polymorphic DNA (RAPD) as genetic markers for determining the origin of interspecific lilac hybrids. Taxon 42: 531–537. [CrossRef][ISI]

McKelvey, S. D.1928.The lilac. Macmillan, New York, NY.

Nei, M.1987.Molecular evolutionary genetics. Columbia University Press, New York, NY.

———, and W.-H. Li.1979.Mathematical model for studying genetic variation in terms of restriction endonucleases. Proceedings of the National Academy of Sciences, USA 76: 5269–5273.

Olmstead, R. G., and J. D. Palmer.1992.A chloroplast DNA phylogeny of the Solanaceae: subfamilial relationships and character evolution. Annals of the Missouri Botanical Garden 79: 346–360. [CrossRef][ISI]

———, and ———.1994.Chloroplast DNA systematics: a review of methods and data analysis. American Journal of Botany 81: 1205–1224. [CrossRef][ISI]

Palmer, J. D.1986.Isolation and structural analysis of chloroplast DNA. Methods in Enzymology 118: 167–186. [ISI]

———, S. R. Downie, J. M. Nugent, P. Brandt, M. Unseld, M. Klein, A. Brennicke, W. Schuster, and T. Borner.1994.The chloroplast and mitochondrial DNAs of Arabidopsis thaliana: conventional genomes in an unconventional plant. In C. Sommerville and E. Meyermowitz [eds.], Arabidopsis, 37–62. Cold Spring Harbor Press, New York, NY.

Parks, C. R., and J. F. Wendel.1990.Molecular divergence between Asian and North American species of Liriodendron (Magnoliaceae) with implications of fossil floras. American Journal of Botany 77: 1243–1256. [CrossRef][ISI]

Pringle, J. S.1977.Interspecific hybridization experiments in Syringa series Villosae (Oleaceae). Baileya 20: 49–91.

———.1981.A review of attempted and reported interseries and intergeneric hybridization in Syringa (Oleaceae). Baileya 21: 101–123.

Rehder, A.1928.Description of the genus and its sections with a key to the species. In S. D. McKelvey [ed.], The lilac, a monograph, 1–14. Macmillan, New York, NY.

———.1945.Notes on some cultivated trees and shrubs. Journal of the Arnold Arboretum 26: 67–78.

———.1977.Manual of cultivated trees and shrubs. Macmillan, New York, NY.

Rieseberg, L. H., and S. J. Brunsfeld.1992.Molecular evidence and plant introgression. In P. S. Soltis, D. E. Soltis, and J. J. Doyle [eds.], Molecular systematics of plants, 151–176. Chapman and Hall, New York, NY.

Sanderson, M. J.1989.Confidence limits on phylogenies: the bootstrap revisited. Cladistics 5: 113–129. [ISI]

Sax, K.1930.Chromosome number and behavior in the genus Syringa. Journal of the Arnold Arboretum 11: 7–14.

———.1945.Lilac species hybrids. Journal of the Arnold Arboretum 26: 79–85.

———.1947.Plant breeding at the Arnold Arboretum. Arnoldia 7: 9–12.

Singh, R. J., and T. Hymowitz.1985a.The genomic relationships among six wild perennial species of the genus Glycine subgenus Glycine Willd. Theoretical and Applied Genetics 71: 221–230. [ISI]

———, and ———.1985b.Intra- and Interspecific hybridization in the genus Glycine, subgenus Glycine Willd.: chromosome pairing and genome relationships. Zeitschrift für Pflanzenzuchtg 95: 289–310.

Skoflek, I.1968.Quarternare Syringa-arten von vertesszollos und monosbel. Acta Botanica Academiae Scientiarum Hungaricae 14: 133–145.

Soltis, D. E., and R. K. Kuzoff.1995.Discordance between nuclear and chloroplast phylogenies in the Heuchera group (Saxifragaceae). Evolution 49: 727–742. [CrossRef][ISI]

———, P. S. Soltis, T. G. Collier, and M. L. Edgerton.1991.Chloroplast DNA variation within and among genera of the Heuchera group: evidence for chloroplast capture and paraphyly. American Journal of Botany 78: 1091–1112. [CrossRef]

Swofford, D. L.1993.PAUP: phylogenetic analysis using parsimony, version 3.1. Illinois Natural History Survey, Champaign, IL.

Taylor, H.1945.Cyto-taxonomy and phylogeny of the Oleaceae. Brittonia 5: 337–369. [CrossRef]

Wendel, J. F.1989.New World tetraploid cottons contain Old World cytoplasm. Proceedings of the National Academy of Sciences, USA 86: 4132–4136. [Abstract/Free Full Text]

———, and V. A. Albert.1992.Phylogenetics of the cotton genus (Gossypium): character-state weighted parsimony analysis of chloroplast-DNA restriction site data and its systematic and biogeographic implications. Systematic Botany 17: 115–143. [CrossRef][ISI]

———, A. Schnabel, and T. Seelanan.1995.Bidirectional interlocus concerted evolution following allopolyploid speciation in cotton (Gossypium). Proceedings of the National Academy of Sciences, USA 92: 280–284. [Abstract/Free Full Text]

Yamazaki, T.1993.Oleaceae. In K. Iwatsuki, T. Yamazaki, D. E. Boufford and H. Ohba [eds.], Flora of Japan 3:122–135. Kodansa Ltd., Tokyo.

This Article Abstract Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted