Am. J. Bot.
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (29)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Plunkett, G. M.
Right arrow Articles by Downie, S. R.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Plunkett, G. M.
Right arrow Articles by Downie, S. R.
Agricola
Right arrow Articles by Plunkett, G. M.
Right arrow Articles by Downie, S. R.
(American Journal of Botany. 1999;86:1014-1026.)
© 1999 Botanical Society of America, Inc.

Major lineages within Apiaceae subfamily Apioideae: a comparison of chloroplast restriction site and DNA sequence data1

Gregory M. Plunkett2 and Stephen R. Downie

Department of Plant Biology, University of Illinois, Urbana, Illinois 61801

Received for publication June 8, 1998. Accepted for publication November 13, 1998.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Traditional sources of taxonomic characters in the large and taxonomically complex subfamily Apioideae (Apiaceae) have been confounding and no classification system of the subfamily has been widely accepted. A restriction site analysis of the chloroplast genome from 78 representatives of Apioideae and related groups provided a data matrix of 990 variable characters (750 of which were potentially parsimony-informative). A comparison of these data to that of three recent DNA sequencing studies of Apioideae (based on ITS, rpoCl intron, and matK sequences) shows that the restriction site analysis provides 2.6–3.6 times more variable characters for a comparable group of taxa. Moreover, levels of divergence appear to be well suited to studies at the subfamilial and tribal levels of Apiaceae. Cladistic and phenetic analyses of the restriction site data yielded trees that are visually congruent to those derived from the other recent molecular studies. On the basis of these comparisons, six lineages and one paraphyletic grade are provisionally recognized as informal groups. These groups can serve as the starting point for future, more intensive studies of the subfamily.

Key Words: Apiaceae • Apioideae • chloroplast genome • restriction site analysis • Umbelliferae


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Apioideae are the largest and best-known subfamily of Apiaceae (= Umbelliferae) and include many familiar edible plants (e.g., carrot, parsnips, parsley, celery, fennel, dill, coriander/cilantro, anise, cumin), as well as several deadly poisons (e.g., poison-hemlock, water-hemlock, fool's-parsley). The subfamily is defined by a suite of easily observed and well-known characters, including pinnately lobed or divided leaves with sheathing petioles, herbaceous stems with hollow internodes, compound-umbellate inflorescences, flowers with pentamerous perianths and androecia, and bicarpellate gynoecia that mature into schizocarpic fruits with two ribbed mericarps. These characters make field recognition of most umbellifers a simple task, but the difficulty in identifying these plants to genus and species is renowned. This contrast reflects a complex history of problems in interpreting the phylogeny of Apioideae, and consequently in producing a satisfactory classification system. Much of the difficulty can be attributed to the near uniformity of most floral characters coupled with repeated examples of apparent parallelisms among other features. Leaf-shape changes provide a good illustration of such variation. Many genera, for example, exhibit interspecific transitions from pinnately lobed or divided leaves with broad leaflets at one extreme to decompound leaves with linear or filiform leaflets at the other extreme (e.g., Ammi, Apium, Ligusticum, Lomatium, Seseli, and Tauschia, among many others); in some cases the leaflets are lost altogether (as in the "rachis-leaf" species of Oenanthe, Oxypolis, and Ptilimnium). Many other morphological, breeding-system, and biochemical characters exhibit similarly confounding parallelisms (e.g., Bell, 1971 ; Harborne, 1971 ; Nielsen, 1971 ).

Fruit morphology and anatomy were traditionally viewed as the most promising sources of taxonomic characters, exhibiting some (but not excessive) variation in features such as fruit shape, the degree and direction of mericarp compression, modifications of the pericarp ribs (e.g., wings or spines), and the shape of mericarp commissural faces. Thus, most traditional classifications of Apiaceae have relied almost exclusively on fruit characters (Koch, 1824 ; Bentham, 1867 ; Boissier, 1872 ; Drude, 1897–1898 ; and Koso-Poljansky, 1916 ; reviewed in Constance, 1971 ; Plunkett, Soltis, and Soltis, 1996b ). Although these systems are now widely regarded as artificial (e.g., Mathias, 1971 ; Theobald, 1971 ; Cronquist, 1982 ; Shneyer et al., 1992 ; Shneyer, Borschtschenko, and Pimenov, 1995 ), the lack of acceptable alternatives has led most students of the family to employ the system proposed by Drude (1897–1898) a century ago in Die Natürlichen Pflanzenfamilian (Table 1). Present-day modifications of this system (e.g., Heywood, 1993 ; Pimenov and Leonov, 1993 ) all retain Drude's basic division of Apiaceae into three subfamilies: Hydrocotyloideae, Saniculoideae, and Apioideae. Hydrocotyloideae and Saniculoideae are much smaller subfamilies (42 genera with ~470 species, and nine genera with ~300 species, respectively; cf. 250–400 genera with 1800–3000 species in Apioideae), and although some questions regarding the relationship of Saniculoideae and Hydrocotyloideae to the apioids persist, recent studies (e.g., Downie et al., 1998 ; Plunkett, Soltis, and Soltis, 1996b , 1997 ) suggest that subfamily Apioideae is monophyletic and that phylogenetic problems in this subfamily can be treated as distinct.


View this table:
[in this window]
[in a new window]
  		Table 1. The subfamilial classification of Apiaceae according to Drude (1897–1898; see also Heywood, 1993); the eight tribes of subfamily Apioideae and the two subtribes of Scandiceae are also provided. Vittae and companion canals are schizogenous oil ducts found in apiaceous fruits (companion canals are associated with vascular bundles; vittae are located between vascular bundles).

 
Providing robust phylogenetic hypotheses is a crucial precursor to erecting stable classification systems. This goal is especially important for Apioideae, which has served as an important system in many evolutionary studies. Most notable among these have been the studies of interactions between umbelliferous host plants and their various insect herbivores (especially the lepidopteran species in the genera Papilio, Depressaria, and Greya; reviewed in Berenbaum, 1983 , 1986 , 1990 ; Thompson, 1986 , 1994 ). To the extent that taxonomy in Apioideae is widely held to be inadequate, especially at tribal and generic levels, the absence of a phylogenetic context among and within umbellifer groups greatly hampers the ability to interpret coevolutionary patterns (see Thompson, 1986 ). Studies of mating systems provide another example where Apioideae have been used as a "model system." Despite frequent descriptions of the subfamily as being bisexual (e.g., Cronquist, 1981 ), many taxa are andromonoecious (Bell, 1971 ; Lovett Doust, 1980 ; Lovett Doust and Harper, 1980 ; Webb, 1981 ,1984 ). It is often assumed that andromonoecy is a derived adaptation resulting from increased selection for dichogamy to promote outcrossing (see Bell, 1971 ; Webb, 1979 ). However, recent studies (Schlessman, Lloyd, and Lowry, 1990 ; Plunkett, Soltis, and Soltis, 1997 ) suggest that andromonoecy may represent the ancestral condition in the entire order Araliales (= Apiaceae plus Araliaceae). Correctly interpreting the evolution and adaptive significance of mating systems in Apioideae requires the correct assessment of polarity, which in turn depends on the presence of well-supported phylogenies. Without such phylogenetic hypotheses, these and other evolutionary studies in Apioideae are largely without context.

Recently, several studies have demonstrated the utility of molecular data in examining evolutionary relationships involving Apiaceae, including studies at the interspecific level (Soltis and Kuzoff, 1993 ; Soltis and Novak, 1997 ), the tribal level (Lee et al., 1997 ), subfamilial level (Downie and Katz-Downie, 1996 ; Downie, Katz-Downie, and Cho, 1996 ; Downie et al., 1998 ), as well as intra- and interfamilial levels (Plunkett, Soltis, and Soltis, 1996a , b , 1997 ). These studies, mostly based on DNA sequence data (from both chloroplast and nuclear markers), have provided insights into the evolutionary history of Apiaceae and hold the promise of producing a framework from which the confusing array of morphological variation can be interpreted. Among the results from these studies are that subfamily Hydrocotyloideae appears to be polyphyletic, but Apioideae and Saniculoideae form monophyletic sister groups. These studies do not, however, support any tribal system of the family, particularly within Apioideae. As a complement to the recent sequencing studies, we undertook an analysis of restriction site data derived from the chloroplast genome of 79 species, with a particular emphasis on subfamily Apioideae. Despite certain limitations, restriction site data confer a number of advantages over sequence data, the most important being that a nearly random sample of the entire chloroplast genome can be surveyed, including data from both rapidly and more slowly evolving sequences (Olmstead and Palmer, 1994 ). Phylogenetic hypotheses based on restriction site data can be examined for areas of congruence and/or conflict with other data sets in the effort to recognize strongly supported evolutionary lineages.

The earlier molecular studies (Downie and Katz-Downie, 1996 , Downie, Katz-Downie, and Cho, 1996 ; Kondo et al., 1996 ; Plunkett, Soltis, and Soltis, 1996b , 1997 ; Downie et al., 1998 ) have clearly demonstrated the problems inherent in most tribal and intergeneric classifications of Apioideae. Data from chloroplast restriction site analysis confirm these results (see below). In an effort to identify evolutionary lineages within Apioideae, the present study seeks to compare results based on different molecular data sets. Given the size of Apioideae (up to 3000 species), it is impractical to build data sets from the entire subfamily for each new study. We hope that the preliminary groupings presented herein will represent starting points for future, more focused studies within Apioideae.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 LITERATURE CITED
 
Fresh or dried leaf tissue was obtained from 79 species (see Table 2) representing Apiaceae (69 species) and the closely related families Araliaceae (eight species) and Pittosporaceae (two species). The two members of Pittosporaceae were included for outgroup comparison on the basis of previous studies (see Chase et al., 1993 ; Plunkett, Soltis, and Soltis, 1996a ). Because relationships within Apioideae were our primary interest, 63 species were sampled from this subfamily; a smaller number of saniculoids (two species) and hydrocotyloids (four species) were included as reference taxa. Among the apioids sampled, seven of the eight tribes originally proposed by Drude (Table 1) were represented. Included from the largest two tribes, Apieae and Peucedaneae, were samples from 32 and 13 species, respectively. From the smaller tribes, there were five species from Smyrnieae, eight from Scandiceae (including the segregate tribe Caucalideae), two from Dauceae, and one species each from Laserpitieae and Coriandreae.


View this table:
[in this window]
[in a new window]
  		Table 2. Voucher or sources and accession numbers for the 79 species of plants used in the present study (tribal classification of subfamily Apioideae based on Drude, 1897–1898).

 
Total DNA was extracted from the leaf tissue using the CTAB method of Doyle and Doyle (1987) and purified using cesium chloride/ethidium bromide gradients (Sambrook, Fritsch, and Maniatis, 1989 ). DNA samples were digested singly with each of the following 14 restriction endonucleases: Ava I, BamH I, Ban II, Bcl I, Bgl II, Cla I, Dra I, Eco O109 I, EcoR I, EcoR V, Hinc II, Hind III, Nci I, and Vsp I. These enzymes recognize 6-bp sequences (except Nci I, which has a 5-bp recognition sequence), cut tobacco cpDNA 40–147 times, and have been used successfully in other plant groups. DNA fragments from each digest were separated electrophoretically using 1.0% agarose gels (in which the bromophenol-blue dye marker was run 10 cm) and then bidirectionally transferred to nylon filters (MSI MagnaCharge, Micron Separations, Westborough, Massachusetts) (Southern, 1975 ). Each of 43 subclones representing the entire tobacco chloroplast genome (described in Olmstead and Palmer [1992] and kindly provided by J. Palmer, Indiana University, Bloomington, Indiana) were labeled with 32P by random priming. The nylon filters were probed with the radiolabeled subclones and then washed in 2xSSC, 0.5% SDS twice for 5 min at room temperature and twice for 60 min at 65°C. Fragment patterns were visualized by autoradiography. Restriction site maps of the entire chloroplast genome were constructed for each of the 14 enzymes. Fragment lengths from each digest were estimated by including two lanes of a size marker (EcoR I/Hind III digested lambda phage DNA) and one lane of tobacco DNA (digested with the enzyme of interest). Expected restriction site maps of the entire tobacco chloroplast genome (Shinozaki et al., 1986 ) were constructed for each enzyme by computer analysis. Because many of the tobacco restrictions sites are conserved among the apioid taxa, a comparison of expected tobacco fragments and observed apioid fragments facilitated map construction.

Phylogenetic analysis of a data matrix based on variable restriction site mutations (available from the authors) was conducted with Wagner parsimony using test version 4.0d63 of PAUP* (D. L. Swofford, Smithsonian Institution, Washington, D.C.). The search options included 1000 replicates (with random addition) of a heuristic search with MULPARS in effect and ACCTRAN optimization. Early trials indicated that the shortest trees were 3038 steps long, but that the analysis was prone to getting stuck on large "islands" (sensu Maddison, 1991 ) of suboptimal trees (3039 steps or longer). For this reason, no more than 500 suboptimal trees were saved per replicate (swapping all saved trees to completion). This search yielded a single island of 84 trees (each of 3038 steps). To test confidence among the nodes of the trees, bootstrap (Felsenstein, 1985 ) and decay (Bremer, 1988 ; Donoghue et al., 1992 ) analyses were carried out. The bootstrap analysis was performed using PAUP*, with 1000 replicates, saving no more than 1000 trees per replicate. To complete the decay analysis, the computer program AutoDecay (Eriksson, 1997 ) was used with PAUP*, following the converse-constraint method (Baum, Sytsma, and Hoch, 1994 ). The data sets were examined for phylogenetic signal using the skewness test (generating the g1 statistic by examining the distribution of 10 000 random trees using the Random Trees function of PAUP*; see Hillis, 1991 ; Huelsenbeck, 1991 ; Hillis and Huelsenbeck, 1992 ; but also Källersjö et al., 1992 ), and a randomization test (the permutation tail probability or PTP test, performed using the permutation function of PAUP* with 1000 replicates of a heuristic search, saving no more than 100 trees per replicate; see Archie, 1989 ; Faith and Cranston, 1991 ). Additionally, PAUP* was used to construct a neighbor-joining tree (for comparison to the parsimony trees) and to generate a distance matrix to examine levels of divergence.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 LITERATURE CITED
 
The chloroplast genomes of apioid taxa are colinear in gene arrangement and readily aligned to tobacco cpDNA. However, one probe (2.1 kb in length, corresponding to the location of trnD, ORF154, psbM, and ORF29 in the large single-copy [LSC] region of tobacco) failed to yield readable fragments, and the fragment patterns from a second probe (2.4 kb long, corresponding to the location of rp116 and rps3 in the LSC of tobacco cpDNA) could not be interpreted without ambiguity. The size of the chloroplast genome among apioids ranges from ~140 to 155 kb. This size variation is attributable to major deletions in one of the inverted repeats (IR) at the boundary of the LSC region. Based on our present analysis, there appear to be at least four distinct IR sizes: ~10 kb, ~17 kb, ~23 kb and ~26 kb. The size of the probes used in this study (ranging from 1.06 to 5.49 kb) prevents a more precise mapping of these deletions (or detection of other, very small insertions/deletions), but this issue will be addressed in a subsequent study based on finer scale probes of the IR-LSC region.

Mapping of restriction sites from 79 taxa in Apiaceae and the closely related families Araliaceae and Pittosporaceae yielded a data matrix of 990 characters, of which 750 were potentially parsimony-informative (240 were found in only a single taxon). Of the total 990 characters, 162 were derived from the small single-copy (SSC) region of the chloroplast genome, 111 from the IRs (scored only once), and 717 from the LSC (Table 3). Divergence values, calculated as mean character difference, ranged from 29.9% (between Scandix and the outgroup taxon Hymenosporum) to 1.0% (between two species of Arracacia and between Anethum and Foeniculum). Within Apiaceae, the range was 27.4% (between Tordylium and the hydrocotyloid Centella) to 1.0%; the range within Apioideae was 22.7% (between Bupleurum and Pastinaca) to 1.0% (Table 4). Among the much smaller sample of taxa from Araliaceae, divergence ranged from 9.5% (between Aralia and Tetrapanax) to 2.4% (between Pseudopanax and Polyscias). In tests for phylogenetic signal, the skewness test yielded a g1 statistic of -0.516, and the permutation analysis yielded a PTP value of 0.001. These results are significant above the 99% confidence level and suggest that the data contain significant amounts of nonrandom structure and differ significantly from randomized data.


View this table:
[in this window]
[in a new window]
  		Table 3. Number, location, and level of homoplasy (HI) for variable restriction sites derived from each of the 14 enzymes used to construct cpDNA maps. Location abbreviations: LSC—large single-copy region of the chloroplast genome; IR—inverted repeat; SSC—small single-copy region (restriction sites from the IR were scored only once).

 

View this table:
[in this window]
[in a new window]
  		Table 4. Comparison of chloroplast restriction site data (present study) to three previously published data sets based on DNA sequences from Apioideae (ITS and rpoC1 intron sequence data, Downie et al., 1998; matK sequence data, Plunkett, Soltis, and Soltis, 1996b); abbreviations: HI = homoplasy index; CI = consistency index of most parsimonious trees; RI = retention index (HI, CI, and RI values are based on data sets excluding uninformative characters). Divergence values are provided for comparison, although sequence divergence is not directly comparable to that of restriction sites.

 
Cladistic analysis of the restrictions site data set yielded 84 most-parsimonious trees of 3038 steps. Excluding uninformative characters, the tree length was 2798, the consistency index (CI) was 0.268 (homoplasy index [HI] of 0.732), and the retention index (RI) was 0.682. The large difference between the CI and RI suggests that the high levels of homoplasy are due in large part to the size of the data matrix (see below). To examine the relative amount of homoplasy from each of the three major regions of the chloroplast genome, characters derived from the other two regions were excluded and the HI was measured against the shortest trees. The SSC exhibited the highest level of homoplasy (0.707), followed by the LSC (0.674); the slowly evolving IR exhibited the least (0.612). A similar comparison was made among the characters derived from each of the 14 enzymes. Characters derived from Dra I had the highest levels of homoplasy (0.779), whereas those from Hinc II showed the least (0.525). A comparison of these values for each enzyme (Table 3) indicates that homoplasy is not necessarily correlated with the number of characters, and some enzymes providing a large number of variable characters (e.g., Nci I, Ban II) have lower homoplasy than several other enzymes yielding many fewer characters (e.g., Dra I, Cla I, Vsp I).

The clades in the strict consensus tree (Fig 1) are labeled with group names that were coined in earlier studies (Plunkett, Soltis, and Soltis, 1996b ; Downie et al., 1998 ). Within the monophyletic Apioideae, the largest clade, the "Angelica group," is supported by a bootstrap (BS) percentage of 99 and a decay index value (DI) of 10. Other clades include the "Aegopodium group" (BS = 59%, DI = 2); the "Apium group" (BS = 39%; DI = 1); the "Daucus group" (BS = 67%; DI = 3); the "Aciphylla" group (BS = 79%; DI = 6); the "Oenanthe" group (BS = 100%; DI = 18); and a basal grade of apioids (Heteromorpha-Anginon and Bupleurum). Outside Apioideae, the two saniculoids form a monophyletic group (BS = 100%; DI = 41), but the hydrocotyloids form a grade of basally branching lineages within Apiaceae. Although the sampling of araliads was small, the restriction site cladogram suggests that the Araliaceae are monophyletic (BS = 100%; DI = 29). The neighbor-joining (NJ) tree reveals the identical seven groups, although the Apium group (rather than the Aegopodium group) is sister to the Angelica group in the NJ tree (Figs. 1,2).



View larger version (49K):
[in this window]
[in a new window]
  		Fig. 1. Strict consensus of the 84 most parsimonious trees resulting from the equally weighted parsimony analysis of cpDNA restriction site data from 79 taxa. Bootstrap percentages are provided above each branch, and decay index values are provided below each branch (preceded by the suffix "d-").

 


View larger version (39K):
[in this window]
[in a new window]
  		Fig. 2. Neighbor-joining tree resulting from the distance analysis of cpDNA restriction site data from 79 taxa.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
LITERATURE CITED
 
The utility of restriction site data
Restriction site analysis continues to be an important source of characters for phylogenetic inference in plant groups (e.g., Cota and Wallace, 1997 ; Luckow, 1997 ; Olmstead and Palmer, 1997 ; Panero and Jansen, 1997 ; Rodriguez and Spooner, 1997 ; Soltis and Novak, 1997 ; Steane et al., 1997 ; Sykes, Christensen, and Peterson, 1997 ), but to a large degree, this method has been supplanted by DNA sequencing studies. Compared to sequencing methods, restriction site analysis requires relatively large amounts of highly purified DNA and involves more laborious laboratory work. In addition, mapping restriction site data is generally more time consuming and less straightforward than reading gene sequence data. Despite these limitations, however, restriction site analysis can provide many more variable characters than most sequencing studies, and by employing different enzymes and different regions of the chloroplast genome, a study of restriction sites can be adjusted to meet the needs of a range of different taxonomic levels (Olmstead and Palmer, 1994 ; Jansen, Wee, and Millie, 1998 ). In the present study, for example, 990 variable characters were scored, of which 750 were potentially parsimony-informative. Compared to three recent sequencing studies of Apioideae (ITS and rpoCl intron sequences—Downie et al., 1998 ; and matK sequences—Plunkett, Soltis and, Soltis, 1996b ), the restriction site analysis provided 2.6–3.6 times the number of potentially informative characters (see Table 4). When comparing the levels of divergence within Apioideae across these studies, restriction site analysis provides a greater range (1.0–22.7%) than the two chloroplast sequences (matK and the rpoCl intron, with ranges of 0.1–9.7 and 0–9.8%, respectively). Conversely, restriction site divergence was lower than that found in ITS (1.7–34.3%), where it became difficult to align ITS sequences from taxa of the "basal grade" (Bupleurum, Anginon, and Heteromorpha) to those of the remaining apioids (Downie and Katz-Downie, 1996 ; Downie et al., 1998 ). When comparing homoplasy levels of the shortest trees from each study (excluding uninformative characters), the restriction site analysis exhibited the greatest homoplasy (HI = 0.732; CI = 0.268). However, the HI (and conversely the CI) are negatively affected by the number of taxa and the number of characters in a data set (see Archie, 1989 ). Given the size of the restriction site matrix (79 taxa and 750 informative characters), it is not surprising that homoplasy appears to be high. The retention index (Archie, 1989 ; Farris 1989 ), which is less sensitive to increases in the number of taxa and characters, is a better measure for comparing data sets of different sizes. The RI for the restriction site study was 0.682, which is comparable to that of the ITS cladogram (0.649), although both of these data sets exhibit more homoplasy than the chloroplast sequence data (0.871 for the rpoC1 intron, and 0.818 for matK; see Table 4). In general, comparisons of various data-set metrics suggest that restriction site data remain a valuable source of characters for phylogeny reconstruction.

Phylogenetic relationships
The taxonomic problems in Apiaceae are pervasive, ranging from species and generic circumscriptions to the relationship of Apiaceae to its "sister family," Araliaceae, to the placement of these families among the other dicot groups. This situation makes both "top-down" and "bottom-up" approaches to systematics equally confounding. Simply put, the problem is "where to jump in." Molecular data are not the panacea of all taxonomic problems, but in troublesome groups like Apiaceae, where the array of "traditional" data is confusing at best, molecular approaches provide the first opportunity of dividing the family into workable units or lineages. Like several previous papers (Downie and Katz-Downie, 1996 ; Downie, Katz-Downie, and Cho, 1996 ; Kondo et al., 1996 ; Plunkett, Soltis, and Soltis, 1996a , b , 1997 ), the present study suggests that subfamily Hydrocotyloideae is not monophyletic. In the strict consensus of the restriction site trees (Fig. 1), the hydrocotyloids form a paraphyletic grade at the base of the Apiaceae clade. Studies with more intensive sampling of hydrocotyloids and araliads further suggest that Hydrocotyloideae may in fact be polyphyletic, with some taxa (notably Hydrocotyle, Centella, and Micropleura) more closely allied to Araliaceae than to the rest of Apiaceae (discussed in Plunkett, Soltis, and Soltis, 1996a , 1997 ). Like other recent studies, chloroplast restriction site data also suggest that Apioideae are a well-supported monophyletic group, sister to a monophyletic Saniculoideae. Given that both traditional concepts and molecular data agree that subfamily Apioideae is "natural" or monophyletic, it seems safe to begin a re-evaluation of Apiaceae at this level.

Four data sets with a broad sampling of apioids are now available: chloroplast restriction sites (present study); nuclear ITS sequences and chloroplast rpoC1 intron sequences (Downie et al., 1998 ); and chloroplast matK sequences (Plunkett, Soltis, and Soltis 1996b ) (see Figs. 1, 3–5; hereafter, these data sets will be abbreviated as the "restriction site," "ITS," "rpoC1," and "matK" studies). Much has been written on the conditions under which data sets can or should be combined (reviewed in de Queiroz, Donoghue, and Kim, 1995 ). Regardless of these issues, the sampling overlap of the four apioid data sets is not at present sufficient enough to warrant construction of a single combined data set. On the other hand, the overlap is not negligible. Of the total 97 genera sampled across these four studies, 80 were included in at least two of the studies and 50 in at least three. Thus, although combining these data sets is premature, visual comparison of cladogram topologies provides a highly congruent picture of relationships within Apioideae.



View larger version (47K):
[in this window]
[in a new window]
  		Fig. 3. Strict consensus of 12 000 trees (length = 730 steps) resulting from the equally weighted parsimony analysis of rpoCl intron sequences from 96 taxa (redrawn from Downie et al., 1998 ); CI excluding uninformative characters = 0.560; RI = 0.871. Bootstrap percentages are provided above each branch.

 
Molecular cladograms consistently reveal six major clades (the Angelica group, the Aegopodium group, the Apium group, the Oenanthe group, the Daucus group, and the Aciphylla group) and one basal, paraphyletic grade (the "basal apioid grade"). The strict consensus tree based on restriction site data (Fig. 1) is labeled with these informal group names, which were originally proposed on the basis of matK data by Plunkett, Soltis, and Soltis (1996b) , and later used by Downie et al. (1998) . Each group is named after a large and familiar genus found within that clade. The use of informal names for lineages of genera provides flexible descriptors that can be used in place of traditional tribal names until a more stable and formal classification system can be erected. Based on comparisons of the four apioid cladograms, we have assigned 90 of the 97 genera in these studies to one of the seven groups (Table 5). Given that many of the larger genera in Apioideae may be paraphyletic (or even polyphyletic), the inclusion of a genus within any informal group may be an oversimplification. Moreover, some genera have been placed in a group on the basis of only a single data set. For these reasons, the groups proposed herein must be viewed as provisional. Appreciating these limitations, we hope the erection of such groups can serve as explicit hypotheses that may be tested at greater length by future studies.


View this table:
[in this window]
[in a new window]
  		Table 5. Provisional groupings of 98 apioid genera based on molecular data. Taxa followed by an asterisk (*) are placed in groups on the basis of only one data set. Abbreviations for data sets: 1—chloroplast DNA restriction site data (present study); 2—rpoC1 intron data (Downie et al., 1998); 3—matK data (Plunkett, Soltis, and Soltis, 1996b); 4—ITS data (Downie et al., 1998); 5—rbcL data (Plunkett, Soltis, and Soltis, 1996a); 6—rbcL data (Kondo et al., 1996). Data-set abbreviations followed by a "?" or "x" (e.g., 1 = ?; 1 = X) indicate that a given data set is equivocal or contradicts the inclusion of a genus in a given group, respectively. For comparison, Drude's (1897–1898) tribes are also provided.

 
The largest of the seven lineages is the Angelica group (see Table 5, Figs. 1, 3–5). Of the four apioid data sets, three (restriction sites, rpoC1, matK) resolve nearly identical complements of taxa within well-supported clades (BS > 90%). In the ITS cladogram, most members from the Angelica group form a single large subclade, but others (e.g., Heracleum, Tordylium, Pastinaca, Cnidium, and Levisticum) are found in two smaller subclades with genera from the Apium group (viz. Conium, Smyrniopsis, Prangos). Branching patterns within the Angelica group vary among the four cladograms, but some groups of taxa are placed together in two or more of the trees. For example, most of the trees suggest a relationship among some of the North American taxa (including Lomatium-Cymopterus-Tauschia-Oreonana, and Zizia-Thaspium-Taenidia-Aletes-Spermolepis), although problems with resolution or sampling make definition of these subclades difficult. All trees do, however, resolve a subclade geographically centered in meso-America, including Arracacia, Rhodosciadium, Coulterophytum, Enantiophylla, and Prionosciadium. Some trees include several other meso-American genera, such as Coaxana, Mathiasella, Dahliophyllum, and Donnellsmithia within this subclade. The Angelica group provides the best evidence of the inadequacy of Drude's intergeneric system, drawing taxa from four of his eight tribes (viz. Apieae, Peucedaneae, Smyrnieae, and Coriandreae). Even so, all but two members of Peucedaneae fall within the Angelica group. Thus, although the dorsally compressed, winged fruits diagnostic of this tribe are homoplastic, there may be a developmental predisposition to these fruit types in the Angelica group.

The clade originally designated as the "Apium group" in the matK study formed two distinct subclades. On the basis of ITS and rpoC1 data, Downie et al. (1998) designated each subclade as a distinct group, the Aegopodium group and a more narrowly defined Apium group. All four data sets reveal nearly identical clades of taxa in the Apium group (sensu stricto), except for the ITS tree (which excludes only Conium, Prangos, Smyrniopsis, and Pimpinella, all found in small subclades with members of the Angelica group). In comparison to Drude's system, the Apium group includes taxa from three tribes: Apieae, Smyrnieae, and Peucedaneae. The resolution of a distinct Aegopodium group is based primarily on matK and ITS data. It includes Aegopodium, Carum, Ciclospermum, Crithmum, Trachyspermum, Falcaria, and Olymposciadium. Of these, only Carum and Crithmum were included in the restriction site data set, but these do form a clade. Intron data from rpoC1 show two distinct clades of taxa from the Aegopodium group (Trachyspermum-Crithmum; and Falcaria-Carum-Aegopodium), but these do not form a monophyletic group. All taxa from the Aegopodium group are from Drude's tribe Apieae, with the exception of Lagoecia. The matK tree suggests that this monotypic saniculoid genus should be transferred to the Aegopodium group in Apioideae, a finding consistent with Koso-Poljanski's (1916) treatment. The transfer of Lagoecia is also supported by cotyledon, pollen, stomatal, and floral-development characters (Cerceau-Larrival, 1962 , 1971 ; Guyot, 1966 , 1971 ; Magin, 1980 ; discussed in Plunkett, Soltis, and Soltis, 1996b ).

The Oenanthe group is a well-supported clade (BS > 90%) resolved by all four analyses. It includes Oenanthe, Perideridia, Sium, Berula, Oxypolis, Cicuta, and probably Neogoezia. The rbcL study of Kondo et al. (1996) also provides evidence for this lineage. Data from matK place Bifora and Shoshonea in the Oenanthe group as well, but in contradiction to the ITS cladogram. In the case of Bifora, different species were sampled for each study (the North American species B. americana was used for matK, whereas the ITS study used the Eurasian B. radians), making it difficult to assess the source of the discrepancy. Future studies including both species will be needed to clarify this issue. Conversely, the single species of the monotypic genus Shoshonea was used in both the ITS and the matK studies and was also included in the rpoC1 data set. The agreement of the ITS and rpoC1 cladograms in placing Shoshonea in the Angelica group (among other western North American endemics) suggests that the matK result may be spurious.

The Daucus group represents all taxa sampled from Drude's tribes Laserpitieae, Dauceae, and Scandiceae (both subtribes Scandicinae and Caucalidinae; see Table 1). Drude's tribal system differed from those proposed earlier by Bentham (1867) and Boissier (1872) , who merged the elements of Drude's Dauceae and subtribe Caucalidinae into a single tribe called Caucalideae (or Caucalineae). This treatment united all taxa with distinctly spiny secondary ribs. As so defined, Caucalideae have been extensively studied by Heywood and colleagues (e.g., Heywood and Dakshini, 1971 ; Heywood, 1973 , 1978 , 1983 ; Jury, 1978 , 1986 ), and have been employed by Pimenov and Leonov (1993) . Bentham and Boissier placed Caucalideae near Laserpitieae, which also has secondary ribs (which are winged rather than armed). Molecular data suggest that all of these groups (Drude's Laserpitieae, Dauceae, and Scandiceae including Caucalideae) represent a single evolutionary lineage, the Daucus group. Within this group, restriction site and ITS data resolve three very similar subclades. One subclade is roughly equivalent to Drude's subtribe Scandicinae, including Scandix, Anthriscus, Chaerophyllum, Myrrhis, and Osmorhiza. A second subclade contains only taxa from Drude's subtribe Caucalidinae: Caucalis, Astrodaucus, Chaetosciadium, Torilis, and Turgenia. The third subclade is drawn from several different tribes and includes Daucus, Cuminum, Laserpitium, Orlaya, and Pseudorlaya.

The Aciphylla group also comprises distinct subclades, one with Aciphylla and Anistome and a second with Lecokia and Smyrnium. Geographically, these subclades are well separated: Lecokia and Smyrnium are native to Eurasia and northern Africa, whereas Aciphylla and Anistome are restricted to New Zealand and Australia. It is likely that sampling additional taxa may serve to bridge the geographic disjunction between these two subclades. The alliance of the Australasian taxa to largely Eurasian genera, however, does indicate that the distinctive apioids of the South Pacific (e.g., Aciphylla, Anistome, Scandia, Gingidia, and Lignocarpa; see Dawson, 1971 ; Dawson and Webb, 1978 ) may not represent ancient relicts but rather may be derived from Eurasian stock. Thus, although the most ancient extant lineages of the order Apiales (Apiaceae and Araliaceae) appear to persist in Australasia (see Plunkett, Soltis, and Soltis, 1996a , 1997 ), the four apioid data sets are unified in suggesting that the basal lineages of Apioideae persist in southern Africa. All molecular data reveal a basal paraphyletic grade comprising Heteromorpha-Anginon and Bupleurum ("basal apioid grade" in Figs. 1, 3–5). Data based on rbcL sequences (Plunkett, Soltis, and Soltis, 1996a ) also support this topology. Both Heteromorpha and Anginon are woody shrubs or small trees restricted in distribution to southern Africa. Bupleurum includes mostly Eurasian herbs, but some species are distinctly woody, and another species is endemic to southern Africa. A more rigorous test of the African origin of Apioideae requires a more intensive study of several other woody African apioids (e.g., Polemannia, Polemanniopsis, Steganotaenia).

Despite the large areas of congruence among the molecular cladograms, several genera are not easily placed. In the case of Ligusticum, it appears that differences are due, at least in part, to the polyphyly of this genus. The restriction site, rpoC1, and rbcL (Kondo et al., 1996 ) data sets all suggest that L. scoticum is allied to the Daucus group; ITS data suggest that it belongs to the Aciphylla group (which in turn is sister to the Daucus group). Two of the data sets (ITS and rbcL) included more than one species of Ligusticum, and these suggest that L. scoticum is not closely related to other species sampled from that genus (L. porteri, L. chuanxiong, L. jeholense, and L. sinense). The other "uncertain taxa" represent genera that are placed in different groups by two or more of the cladograms. For example, Arafoe is placed in the Angelica group on the basis of rpoC1 data, but is placed with Pimpinella (of the Apium group) by the ITS tree. Finally, some genera (e.g., Komarovia, Physospermum, and Conioselinum) form isolated lineages that are difficult to assign to any of the seven groups. Sampling additional genera may help to stabilize the placement of these genera. Alternatively, these taxa may belong to groups as yet undescribed.

In broader terms, the four apioid cladograms suggest that the Angelica group, the Apium group, and the Aegopodium group form a single large clade (the apioid "superclade"). Relationships between the superclade and the remaining three monophyletic groups (the Oenanthe group, the Daucus group, and the Aciphylla group) are less clear. Restriction sites and matK data suggest that the Daucus group is sister to this superclade (Figs. 1, 4), whereas the placement of the superclade in cladograms based on rpoC1 intron and ITS data is equivocal (Fig. 5). All data sets confirm that Heteromorpha, Anginon, and Bupleurum occupy basally branching positions within Apioideae.



View larger version (54K):
[in this window]
[in a new window]
  		Fig. 4. Strict consensus of 902 trees (length = 709) resulting from the equally weighted parsimony analysis of matK sequences from 53 taxa (redrawn from Plunkett, Soltis, and Soltis, 1996b ); CI excluding uninformative characters = 0.570; RI = 0.818. Bootstrap percentages are provided below each branch, and decay index values are provided above each branch (preceded by the suffix "d-").

 


View larger version (46K):
[in this window]
[in a new window]
  		Fig. 5. Strict consensus of 12 000 trees (length = 2107 steps) resulting from the equally weighted parsimony analysis of nuclear ITS sequences from 95 taxa (redrawn from Downie et al., 1998 ); CI excluding uninformative characters = 0.314; RI = 0.649. Bootstrap percentages are provided above each branch.

 
All of the major groups (excluding the basal grade) are monophyletic in at least three of the four apioid cladograms (Figs. 1, 3–5). Of these, the Angelica group, the Oenanthe group, and the Daucus group are each supported by bootstraps of >90% in at least three of the four studies (Table 6). Conversely, the bootstrap support for the Aciphylla group, the Apium group, and the Aegopodium group are low to moderate (ranging from 38 to 79%, 39 to 63%, and 59 to 67%, respectively). Despite such low internal support, clades with nearly identical complements of taxa are resolved by at least three of the four studies. Comparing cladograms from different studies provides an estimate of external support not readily available in studies based on a single source of data. Although three of the four apioid data sets are derived from a single inheritance unit (the chloroplast genome), these represent three different types of data samplings (restrictions sites, intron sequences, and gene sequences). Thus, the four data sets represent not only two independent sources of characters (nuclear and chloroplast DNA), but also four independent samplings. In areas where the four cladograms are largely congruent, we can infer that the data sets from which they were derived contain the same phylogenetic signal. Further, a comparison of the general topologies of each tree and of bootstrap support for individual clade suggests that none of the four data sets excels in its ability to resolve congruent and well-supported clades. These comparisons agree with recent assessments of other groups that multiple data sets will be required before a stable picture of relationships can be discerned. Within Apioideae, we hope that the present grouping of genera into informal groupings can serve as a basic framework in which the complex patterns of morphology, anatomy, biochemistry, and molecular characters can be reinterpreted, and from which a new and more satisfactory classification system can be erected.


View this table:
[in this window]
[in a new window]
  		Table 6. A comparison of bootstrap percentages for the six monophyletic groups resolved by at least three of the four apioid data sets (chloroplast DNA restriction site data—present study; rpoC1 intron data—Downie et al., 1998; matK data—Plunkett, Soltis, and Soltis, 1996b; ITS data—Downie et al., 1998). The apioid "superclade" comprises a monophyletic group of the Angelica group, the Apium group, and the Aegopodium group. Symbols: * = not a monophyletic group in a given cladogram (for the ITS tree, values of roughly equivalent subclades are provided in parentheses); {dagger} = only a single taxon included from this group.

 


View this table:
[in this window]
[in a new window]
  		Table 2. Continued.

 

View this table:
[in this window]
[in a new window]
  		Table 5. Continued.

 

   FOOTNOTES
 
1 The authors thank L. Constance, J. Lahham, B. Lee, J. Lyons-Weiler, and many botanical gardens (cited in the text) for kindly providing leaf and/or seed material; and V. Houpis, R. Patel, C. Germann, and M. Burke for assistance in the laboratory. This work was supported by a grant from the National Science Foundation (DEB-9407712). Back

2 Author for correspondence, current address: Department of Biology Virginia Commonwealth University, Richmond, VA 23284-2012. Back


   LITERATURE CITED
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Archie, J. W. 1989 A randomization test for phylogenetic information in systematic data. Systematic Zoology 38: 239–252.[CrossRef]

Baum, D. A., K. J. Sytsma, and P. C. Hoch. 1994 A phylogenetic analysis of Epilobium (Onagraceae) based on nuclear ribosomal DNA sequences. Systematic Botany 19: 363–388.[CrossRef][ISI]

Bell, C. R. 1971 Breeding systems and floral biology of the Umbelliferae, or evidence for specialization in unspecialized flowers. In V. H. Heywood [ed.], The biology and chemistry of the Umbelliferae, Supplement 1. Botanical Journal of the Linnean Society 64: 93–108.

Bentham, G. 1867 Umbelliferae. In G. Bentham and J. D. Hooker [eds.], Genera Plantarum 1: 859–931.

Berenbaum, M. R. 1983 Coumarins and caterpillars: a case for coevolution. Evolution 37: 163–179.[CrossRef][ISI]

———. 1986 Constrains on chemical coevolution: wild parsnips and the parsnip webworm. Evolution 40: 1215–1228.[CrossRef][ISI]

———. 1990 Evolution of specialization in insect-umbellifer associations. Annual Review of Entomology 35: 319–343.[CrossRef][ISI]

Boissier, E. 1872 Umbelliferae. Flora Orientalis 2: 892–1091.

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

Cerceau-Larrival, M. T. 1962 Plantules et pollens d'Ombelliféres: leur intérêt systématique et phylogénetique. Mémoires du Muséum National d'Histoire Naturelle, Série B. Botanique 14: 1–164.

———. 1971 Morphologie pollinique et corrélations phylogénétiques chez les Ombelliféres. In V. H. Heywood [ed.], The biology and chemistry of the Umbelliferae, Supplement 1, Botanical Journal of the Linnean Society 64: 109–135.

Chase, M. W., et al. 1993 Phylogenetics of seed plants: an analysis of nucleotide sequences from the plastid gene rbcL. Annals of the Missouri Botanical Garden 80: 528–580.[CrossRef][ISI]

Constance, L. 1971 History of the classification of Umbelliferae (Apiaceae). In V. H. Heywood [ed.], The biology and chemistry of the Umbelliferae, Supplement 1. Botanical Journal of the Linnean Society 64: 1–8.

Cota, J. H., and R. S. Wallace. 1997 Chloroplast DNA evidence for divergence in Ferocactus and its relationships to North American columnar cacti (Cactaceae: Cactoideae). Systematic Botany 22: 529–542.[CrossRef][ISI]

Cronquist, A. 1981 An integrated system of classification of flowering plants. Columbia University Press, New York, NY.

———. 1982 Reduction of Pseudotaenidia to Taenidia (Apiaceae). Brittonia 34: 365–367.[CrossRef][ISI]

Dawson, J. W. 1971 Relationships of the New Zealand Umbelliferae. In V. H. Heywood, [ed.], The biology and chemistry of the Umbelliferae, Supplement 1, Botanical Journal of the Linnean Society 64: 43–62.

———, and C. J. Webb. 1978 Generic problems in Australiasian Apioideae (Umbelliferae). In A. M. Cauwet and J. Carbonnier [eds.], Actes du 2e Symposium International sur les Ombellifères, Contributions Pluridesciplinaires à la Systématique, 21–32. Missouri Botanical Garden, St. Louis, MO.

de Queiroz, A., M. J. Donoghue, and J. Kim 1995 Separate versus combined analysis of phylogenetic evidence. Annual Review of Ecology and Systematics 26: 657–681.[ISI]

Donoghue, M. J, R. G. Olmstead, J. F. Smith, and J. D. Palmer. 1992 Phylogenetic relationships of Dipsacales based on rbcL sequences. Annals of the Missouri Botanical Garden 79: 333–345.

Downie, S. R., and D. S. Katz-Downie. 1996 A molecular phylogeny of Apiaceae subfamily Apioideae: evidence from nuclear ribosomal DNA internal transcribed spacer sequences. American Journal of Botany 83: 234–251.[CrossRef][ISI]

———, ———, and K.-J. Cho. 1996 Phylogenetic analysis of Apiaceae subfamily Apioideae using nucleotide sequences from the chloroplast rpoC1 intron. Molecular Phylogenetics and Evolution 6: 1–18.[CrossRef][ISI][Medline]

———, S. Ramanath, D. S. Katz-Downie, and E. Llanas. 1998 Molecular systematics of Apiaceae subfamily Apioideae: phylogenetic analyses of nuclear ribosomal DNA internal transcribed spacer and plastid rpoC1 intron sequences. American Journal of Botany 85: 563–591.[Abstract]

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.

Drude, C. G. O. 1897–1898 Umbelliferae. In A. Engler and K. Prantl [eds.], Die natürlichen Planzenfamilian 38: 63–250.

Eriksson, T. 1997 AutoDecay version 2.9.9 (hypercard-stack computer program distributed by the author). Botaniska Institutionem, Stockholm University, Stockholm, Sweden.

Faith, D., and P. Cranston. 1991 Could a cladogram this short have arisen by chance alone? Cladistics 7: 1–28.

Farris, J. S. 1989 The retention index and the rescaled consistency index. Cladistics 5: 417–419.[ISI]

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

Guyot, M. 1966 Les stomates des Ombelliferes. Bulletin de la Societé Botanique de France 113: 244–273.

———. 1971 Phylogenetic and systematic value of stomata of the Umbelliferae. In V. H. Heywood, [ed.], The biology and chemistry of the Umbelliferae, Supplement 1. Botanical Journal of the Linnean Society 64: 199–214.

Harborne, J. B. 1971 Flavonoid and phenylpropanoid patterns in the Umbelliferae. In V. H. Heywood [ed.], The biology and chemistry of the Umbelliferae, Supplement 1. Botanical Journal of the Linnean Society 64: 293–314.

Heywood, V. H. 1973 The taxonomic position of Agrocharis Hochst. and allied genera. Notes from the Royal Botanic Garden, Edinburgh 32: 211–215.

———. 1978 Multivariate taxonomic synthesis of the tribe Caucalideae. In A. M. Cauwet and J. Carbonnier [eds.], Actes du 2e Symposium International sur les Ombellifères, Contributions Pluridesciplinaires à la Systématique, 727–736. Missouri Botanical Garden, St. Louis, MO.

———. 1983 Relationships and evolution in the Daucus carota complex. Israel Journal of Botany 32: 51–65.

———. 1986 The Umbelliferae—an impossible family? Symbolae Botanicae Upsalienses 26: 73–80.

———. 1993 Flowering plant families of the world. Oxford University Press, New York, NY.

———, and K. M. M. Dakshini. 1971 Fruit structure in the Umbelliferae-Caucalideae. In V. H. Heywood [ed.], The biology and chemistry of the Umbelliferae, Supplement 1. Botanical Journal of the Linnean Society 64: 217–232.

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

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

Huelsenbeck, J. P. 1991 Tree-length distribution skewness: an indicator of phylogenetic information. Systematic Zoology 40: 257–270.[CrossRef]

Jansen, R. K., J. L. Wee, and D. Millie. 1998 Comparative utility of chloroplast DNA restriction site and DNA sequence data for phylogenetic studies in plants. In D. E. Soltis, P. S. Soltis, and J. J. Doyle [eds.], Molecular systematics of plants II, 87–100. Kluwer Academic Press, Boston, MA.

Jury, S. 1978 Tuberculate fruit in the Umbelliferae. In A. M. Cauwet and J. Carbonnier [eds.], Actes du 2e Symposium International sur les Ombellifères, Contributions Pluridesciplinaires à la Systématique, 149–160. Missouri Botanical Garden, St. Louis, MO.

———. 1986 Fruit and leaf variation in the African species of the Umbelliferae tribe Caucalideae. Symbolae Botanicae Upsalienses 26: 181–188.

Källersjö, M., J. S. Farris, A. G. Kluge, and C. Bult. 1992 Skewness and permutation. Cladistics 8: 275–287.

Koch, W. D. J. 1824 Generum tribuumque plantarum Umbelliferarum nova dispositio. Nova Acta Academiae Caesareae Leopoldino Carolinae Germanicae Naturae Curiosorum 12: 55–156.

Kondo, K., S. Terabayashi, M. Okada, C. Yuan, and S. He. 1996 Phylogenetic relationship of medicinally important Cnidium officinale and Japanese Apiaceae based on rbcL sequences. Journal of Plant Research 109: 21–27.

Koso-Poljansky, B. 1916 Sciadophytorum systematis lineamenta. Bulletin de la Société Impériale des Naturalistes de Moscou 29: 93–221.

Lee, B., S. Ramanath, M. Choi, and S. R. Downie. 1997 A molecular phylogeny of Umbelliferae tribe Caucalideae: evidence from nuclear ribosomal DNA internal transcribed spacer (ITS) sequences. American Journal of Botany 84: 208 (Abstract).[Abstract]

Lovett Doust, J. 1980 Floral sex ratios in andromonoecious Umbelliferae. New Phytologist 85: 265–273.[CrossRef][ISI]

———, and J. L. Harper. 1980 The resource costs of gender and maternal support in an andromonoecious umbellifer, Smyrnium olustrum L. New Phytologist 85: 251–264.[CrossRef][ISI]

Luckow, M. 1997 Generic relationships in the Dichrostachys group (Leguminosae: Mimosoideae): evidence from chloroplast DNA restriction sites and morphology. Systematic Botany 22: 189–199.

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

Magin, N. 1980 Eine blütenmorphologische Analyse der Lagoecieae (Apiaceae). Plant Systematics and Evolution 133: 239–259.[CrossRef][ISI]

Mathias, M. E. 1971 Systematic survey of New World Umbelliferae. In V. H. Heywood [ed.], The biology and chemistry of the Umbelliferae, Supplement 1. Botanical Journal of the Linnean Society 64: 13–30.

Nielsen, B. E. 1971 Coumarin patterns in the Umbelliferae. In V. H. Heywood [ed.], The biology and chemistry of the Umbelliferae, Supplement 1. Botanical Journal of the Linnean Society 64: 325–336.

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]

———, and ———. 1997 implications for the phylogeny, classification, and biogeography of Solanum from cpDNA restriction site variation. Systematic Botany 22: 19–29.

Panero, J. L., and R. K. Jansen. 1997 Chloroplast DNA restriction site study of Verbesina (Asteraceae: Heliantheae) American Journal of Botany 84: 382–392.[Abstract]

Pimenov, M. G., and M. V. Leonov. 1993 The genera of the Umbelliferae: a nomenclator. Royal Botanic Gardens, Kew, UK.

Plunkett, G. M., D. E. Soltis, and P. S. Soltis. 1996a Higher level relationships of Apiales (Apiaceae and Araliaceae) based on rbcL sequences. American Journal of Botany 83: 499–515.[CrossRef][ISI]

———, ———, and ———. 1996b Evolutionary patterns in Apiaceae: inferences based on matK sequence data. Systematic Botany 21: 477–495.[CrossRef][ISI]

———, ———, and ———. 1997 Clarification of the relationship between Apiaceae and Araliaceae based on matK and rbcL sequence data. American Journal of Botany 84: 565–580.[Abstract]

Rodriguez, A., and D. M. Spooner. 1997 Chloroplast DNA analysis of Solanum bulbocastanum and S. cardiophyllum, and evidence for the distinctiveness of S. cardiophyllum subsp. ehrenbergii (sect. Petota). Systematic Botany 22: 31–43.

Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989 Molecular cloning, a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Plainview, NY.

Schlessman, M. A., D. G. Lloyd, and P. P. Lowry. 1990 Evolution of sexual systems in New Caledonian Araliaceae. Memoirs of the New York Botanical Garden 55: 105–117.

Shinozaki, K., et al. 1986 The complete nucleotide sequence of the tobacco chloroplast genome: its gene organization and expression. EMBO Journal 5: 2043–2049.[ISI][Medline]

Shneyer, V. S., G. P. Borschtschenko, and M. G. Pimenov 1995 Immunochemical appraisal of relationships within the tribe Peucedaneae (Apiaceae). Plant Systematics and Evolution 198: 1–16.[CrossRef][ISI]

———, ———, ———, and M. V. Leonov. 1992 The tribe Smyrnieae (Umbelliferae) in the light of serotaxonomical analysis. Plant Systematics and Evolution 182: 135–148.[CrossRef][ISI]

Soltis, P. S., and R. K. Kuzoff. 1993 ITS sequence variation within and among populations of Lomatium grayi and L. laevigatum (Umbelliferae). Molecular Phylogenetics and Evolution 2: 166–170.[CrossRef][Medline]

———, and S. J. Novak. 1997 Polyphyl