Phylogenetic Analysis of Cpdna Restriction Sites and Rpsl6 Intron Sequences Reveals Relationships Among Apiaceae Tribes Caucalideae, Scandiceae and Related Taxa

Plant Syst. Evol. 221:35-60 (2000) Plant Systematics and Evolution © Springer-Verlag 2000 Printed in Austria

Phylogenetic analysis of cpDNA restriction sites and rpsl6 intron sequences reveals relationships among Apiaceae tribes Caucalideae, Scandiceae and related taxa

B.-Y. Lee and S. R. Downie

Department of Plant Biology, University of Illinois at Urbana-Champaign, Urbana, Illinois, USA

Received February 14, 1999 Accepted August 10, 1999

Abstract. Evolutionary relationships among mem- within Torilis, and is recognized as Torilis tricho- bers of Apiaceae (Umbelliferae) tribe Caucalideae sperma (L.) Spreng. Spreng. and related taxa were inferred from maximum parsimony analyses of chloroplast DNA Key words: Caucalideae, Scandiceae, Apiaceae, restriction sites and rpsl6 intron sequences and the Umbelliferae, chloroplast DNA restriction sites, results compared to an existing phylogeny for the rpsl6 intron. group based on nuclear ribosomal DNA internal transcribed spacer sequences. While these three data sets were not similar in size or composition, Introduction the relationships among the shared taxa, with few Tribe Caucalideae Spreng., as defined by exceptions, were concordant. Three major lineages Bentham (1867) and later Boissier (1872), are recognized, coinciding with the previously delimited Scandiceae subtribes Daucinae Dumort. contains practically all of those species of (Agrocharis, Ammodaucus, Cuminum, Daucus, Apiaceae (Umbelliferae) that have spines, Orlaya, Pachyctenium, Pseudorlaya), Torilidinae hooks, tubercles, or bristly hairs on the Dumort. (Astrodaucus, Caucalis, Glochidotheca, primary and/or secondary (vallecular) ridges Lisaea, Szovitsia, Torilis, Turgenia, Yabea), and of their fruits. Uniquely in this group, the Scandicinae Tausch (Anthriscus, Kozlovia, Myrrhis, secondary ridges are often more strongly Osmorhiza, Scandix). Included in Daucinae is developed than the primary. These plants are representation from tribe Laserpitieae (Laser, Lase- distributed throughout Europe, the Mediter- rpitium, Melanoselinum, Monizia, Polylophium). ranean region, and southwestern and central Daucinae and Torilidinae arise as sister taxa in Asia, with a few outlying members in North the chloroplast DNA-based phylogenies, whereas and South America and Australia. Of the 21 in the ITS trees relationships among the three genera and 68 species recognized in the tribe major lineages are unresolved. Unexpectedly, three species of Ferula ally with Daucinae and Torili- (Heywood and Jury in Heywood 1982c), dinae. The position of Artedia is equivocal, occur- Daucus is the largest genus with 21 species ring either sister to Daucinae in the ITS trees, within followed by Torilis with 10 species. Torilidinae in the intron trees, or sister to Torilid- In the most widely used revision of the inae upon analysis of combined ITS and intron family, Drude (1897 1898) redistributed these data. Chaetosciadium trichospermum emerges spiny-fruited plants between his divergent 36 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae

Scandiceae subtribe Caucalidinae and tribe the multidisciplinary approaches used to ana- Dauceae. Drude believed that members of lyse these data, fundamental disagreements tribe Dauceae (e.g. Daucus), having spines on still exist regarding the proper circumscription their prominent secondary fruit ridges, are of tribe Caucalideae, the relationships among allied to plants in tribe Laserpitieae (e.g. its members, and the delimitation of certain Laserpitium and Polylophium), whose members genera. Moreover, the relationship between have fruits without spines but with primary Heywood's tribes Caucalideae and Scandiceae and prominent secondary ridges. In contrast, is also quite unclear. the genera of Scandiceae subtribe Caucalidinae In all phylogenetic analyses of molecular (e.g. Caucalis, Orlaya, and Torilis) are linked data to date, whether derived from DNA (by the common possession of calcium oxalate sequences of chloroplast introns or genes, crystals in the parenchyma cells surrounding chloroplast DNA (cpDNA) restriction sites, the carpophore) to those in Scandiceae sub- or nuclear ribosomal DNA (rDNA) internal tribe Scandicinae (e.g. Anthriscus and Scan- transcribed spacer (ITS) sequences, a close dix), the latter subtribe lacking both secondary relationship between Apiaceae tribes Cauca- ridges and spines. Drude assumed that the lideae and Scandiceae is apparent (Downie secondary spinose ridges characteristic of some et al. 1996, 1998; Downie and Katz-Downie Caucalidinae had evolved independently from 1996; Katz-Downie et al. 1999; Plunkett et al. those in Dauceae. Other treatments exist for 1996; Plunkett and Downie 1999). However, these plants, such as those proposed by Cale- while some trees show that Caucalideae and stani (1905), Koso-Poljansky (1916, 1917), and Scandiceae are monophyletic sister taxa, oth- Cerceau-Larrival (1962, 1965), but have not ers indicate that Caucalideae is paraphyletic gained wide acceptance. with Scandiceae nested within. Alternatively, As a result of two international symposia the marK study of Plunkett et al. (1996) (Heywood 1971a, Cauwet-Marc and Carbon- shows a paraphyletic Scandiceae with includ- nier 1982) and a major cooperative research ed Caucalideae. Because the purpose of each program (reviewed in Heywood 1982a, c), the of these studies was not to resolve the spiny-fruited umbellifers have received much intergeneric relationships within tribes Cau- systematic attention. The multidisciplinary re- calideae and Scandiceae but rather to infer the search undertaken, incorporating results from higher-level groupings within the family, the the fields of scanning electron microscopy, number of taxa sampled from each tribe was biochemical systematics, cytology, and numer- small. ical taxonomy, culminated in changes to the We have recently addressed issues of rela- prevailing taxonomy. Here Drude's Dauceae tionship among the spiny-fruited umbellifers and Scandiceae subtribe Caucalidinae were and related taxa by carrying out, with expand- reunited as tribe Caucalideae (Heywood ed sampling, phylogenetic analyses of nuclear 1968a, 1971c; Crowden et al. 1969; McNeill rDNA ITS sequences (Lee and Downie 1999). et al. 1969; Heywood and Dakshini 1971; Fifty-eight accessions, representing 18 of 21 Harborne and Williams 1972; Williams and genera of Caucalideae (Aphanopleura, Psam- Harborne 1972), and Drude's Scandiceae sub- mogeton, and the rare, monotypic Angoseseli tribe Scandicinae was treated at the tribal level, were not considered) and putatively allied taxa Scandiceae Spreng. (Heywood 1971b). Of the from Heywood's Scandiceae and Drude's 21 genera provisionally recognized in Caucal- tribes Apieae, Laserpitieae, and Smyrnieae, ideae (Heywood and Jury in Heywood 1982c), were examined and analyzed using maximum two (Aphanopleura and Psammogeton) have parsimony, maximum likelihood, and neigh- been excluded from the group (Pimenov and bor-joining methods. The results of the max- Leonov 1993, Katz-Downie et al. 1999). How- imum parsimony analysis are presented in ever, despite the wealth of available data and Fig. 1. Here three major lineages are recog- B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae 37 nized, designated as Scandiceae subtribes Materials and methods Daucinae Dumort. (1827), Torilidinae Terminal taxa. The 52 accessions considered in this Dumort. (1827), and Scandicinae Tausch study, with corresponding source and voucher (1834; Downie et al. 2000). These taxa coincide information, are listed in Table 1. Thirty-two with the previously delimited "Daucus," accessions were included in the mapping of cpDNA "Torilis," and "Scandix" subgroups (Lee and restriction sites and 34 accessions were sequenced Downie 1999) of the "Daucus" clade (Plunkett for the chloroplast rpsl6 intron, with 14 accessions et al. 1996, Downie et al. 1998, Plunkett and common to both analyses. While each data set was Downie 1999). Subtribes Daucinae and Tori- approximately similar in size, 14 genera were lidinae coincide approximately with Hey- included in the restriction site study (where em- wood's (1982c) tribe Caucalideae but with the phasis on sampling was placed on Daucus and inclusion of four genera of Laserpitieae and Torilis), whereas 27 genera were included in the intron study (where emphasis on sampling was the exclusion of Kozlovia. placed on Laserpitieae and Ferula). The genus We now continue our investigations of Ferula, treated in Drude's tribe Peucedaneae and Caucalideae phylogeny by incorporating data thus considered distantly related to the spiny- from the chloroplast genome. Congruence of fruited umbetlifers, allies with Torilis in the ITS relationship from independent lines of evi- study of Valiejo-Roman et al. (1998) and, as such, dence is necessary in order to examine the was included in our investigation. These 52 acces- robustness of earlier phylogenetic hypotheses sions represent 15 of the 21 genera recognized most and to identify discrepant organismal and recently in Caucalideae (Heywood and Jury in gene phylogenies. Here we use cladistic anal- Heywood 1982c). The six remaining genera were ysis of cpDNA restriction sites and chloro- not included due to lack of sufficient material or, as plast rps16 intron sequences to infer historical in the case of Aphanopleura and Psammogeton, relationships. The latter was chosen because because previous studies had supported their removal from the tribe (Pimenov and Leonov of its potential for variation, the ease by 1993, Katz-Downie et al. 1999). Material from which the region can be isolated from her- the rare, monotypic genus Angoseseli has yet to be barium material, and the success others have included in any molecular systematic study. Addi- had in using this locus for phylogenetic tionally, we included representation fi'om Hey- inference in other groups at comparable wood's Scandiceae and Drude's (1897-1898) tribes taxonomic levels (Lid6n et al. 1997, Oxelman Apieae, Laserpitieae, and Smyrnieae, based on the et al. 1997, Downie and Katz-Downie 1999). results of our higher-level molecular phylogenetic Our objectives are to: (1) provide a phyloge- analyses of the family (Downie et al. 1998) and our netic estimate for the group using evidence recent ITS studies of Caucalideae and relatives from cpDNA and to identify well-supported (Katz-Downie et al. 1999, Lee and Downie 1999). monophyletic groups; (2) compare the utility Smyrnium olusatrum (Smyrnieae) was used to root of cpDNA restriction sites and rpsl6 intron the trees. sequences in resolving relationships; and (3) DNA extraction. Total genomic DNA was compare the phylogenetic hypothesis obtained extracted from either fresh leaf or herbarium- preserved tissue using the modified CTAB proce- to that inferred using nuclear rDNA ITS dure of Doyle and Doyle (1987). For some taxa, the sequences. Subsequent studies (manuscripts DNA was purified further by centrifugation to currently in preparation) deal with the cladis- equilibrium in cesium chloride/ethidium bromide tic analysis of morphological data, the inter- gradients. pretation of cytological, palynological, Chloroplast DNA restriction site analysis. anatomical, morphological, and chemical Approximately 1 ~tg of total genomic DNA was character evolution in the group in light of digested singly with each of the following 14 its inferred evolutionary history, and the restriction enzymes (all recognizing 6-bp sequences delimitation of genera based on these mor- except for NciI which recognizes 5 bp sequences) phological and molecular data. according to the manufacturers' instructions: AvaI, 38 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae

Daucus maximus m IO0 Daucus carota subsp, carota Daucus carota subsp, sativus 73 >_2 >2 i 66 I--~ Daucus carota subsp, halophilus >2 1 L---~ Daucus carota subsp, gummifer Pseudorlaya pumila 93 Daucus aureus >2 Daucus muricatus Pachyctenium mirabile Daucus crinitus 97 66 100. Daucus bicolor subsp, broteri >_2 Daucus bicolor subsp, bicolor Daucus pusillus 100~ Daucus montanus 98 >_2~ Daucus durieua 96 O3 >_2 >_2 74 Agrocharis pedunculata t- 100 1 I Agrocharis incognita °mo 91 >2 Agrocharis melanantha _>2 Laserpitium hispidum E) 93 Orlaya daucoides 100 >21 Orlaya daucorlaya 74 >2 Orlaya grandiflora 1 44 Cuminum cyminum 1 >_2" Cuminum setifolium 51 47 Ammodaucus leucotrichus 1 24 1 69 Laser trilobum 1 >_2 I Polylophium panjutinfi Laserpitium slier Artedia squamata Astrodaucus orientalis 63 Glochidotheca foeniculacea 1 Szovitsia callicarpa 96 Torilis arvensis subsp, arvensis >2 Torilis arvensis subsp, purpurea Torilis tenella 95 Torilis scabra 85 >_2 Torilis leptophylla O3 >2 Torilis nodosa ¢- Chaetosciadium trichospermum Torilis elongata 0 Yabea microcarpa I-- Caucalis platycarpos Turgenia latifolia _>28510122_ ~ >_2 I Lisaea strigosa > 2 66 Lisaea papyracea Lisaea heterocarpa 57 Anthriscus caucalis 1 • Anthriscus cerefolium O3 Myrrhis odorata .~_ KozIovia paleacea .o. 1O0 z2[ Osmorhiza Iongistylis t- >_2 100 Scandix pecten-veneris o >.2 I Scandix balansae 03 Smymium olusatrum Lecokia cretica Ligusticum scoticum Aciphylla subflabellata Aciphylla squarrosa B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae 39

BamHI, BanII, BglII, ClaI, DraI, EcoOl09I, HindIII), additional finer-scale hybridization EcoRI, EcoRV, HincII, HindIII, NciI, XbaI, and probes from tobacco cpDNA were used in order XhoI. DNA digests were separated electrophoret- to construct detailed gene maps and to survey for ically in 1.0% agarose gels and bidirectionally specific genes and introns (probes and rationale transferred to Magna Charge nylon membranes described in Downie and Palmer 1992). Fragment (Micron Separations, Inc., Westborough, MA) for sizes were estimated by the inclusion of size filter hybridizations to 32p radiolabeled tobacco markers by combining equimolar mixtures of probes. Thirty-five probes (obtained from J. Palmer, phage lambda DNA digested with EcoRI and Indiana University, Bloomington and described in HindIII and with HindIII alone. One lane of Olmstead and Palmer 1992), covering both the LSC tobacco cpDNA, digested with the enzyme of (large single-copy) and SSC (small single-copy) interest, was also included to facilitate mapping. regions of the chloroplast genome, were used. The A matrix of binary data, representing restriction evolutionary conservatism of restriction sites in the site presence (1) or absence (0), was constructed. inverted repeat region of Apiaceae cpDNAs, as For a few taxa and enzymes (especially XbaI), noted in a concurrent but higher-level study of missing data were scored as "?," a result of either Apiaceae phylogeny (Plunkett and Downie 1999), partial- or non-digestion of DNA. suggested that this region would not supply much Amplification and sequencing of the chloroplast phylogenetic information and, therefore, was not rpsl6 intron. For 34 accessions, a region containing considered. Prehybridization (2-6 hours) and hy- the complete rpsl6 intron and about half of its bridization (24-36 hours) were carried out at 65 °C flanking 3' exon was PCR-amplified using primers in 2× SSC, 0.5% SDS, and 0.25% nonfat dry milk. "5' exon rpsl6" (AAACGATGTGGNAGNAAR- Washing of membranes and autoradiography were CA) and "3' exon rpsl6" (CCTGTAGGYTGNG- as described in Olmstead and Palmer (1992). CNCCYTT) in an equimolar ratio (primers written Individual nylon membranes were reused up to 17 5' to 3'). In tobacco cpDNA, the rpsl6 intron is times by stripping the hybridized probes with 860 bp in size (Shinozaki et al. 1986). Each set of boiling strip solution (0.1x SSC) prior to subse- PCR amplifications was monitored by the inclusion quent prehybridization. Restriction site maps of the of positive (tobacco cpDNA) and negative (no LSC and SSC regions were constructed for each template) controls. Details of primer design, the accession for each of the 14 enzymes. For one PCR-amplification reactions, the DNA purification taxon in each of the three recently recognized and automated DNA sequencing strategies used, subtribes (Daucus carota subsp, sativus, Torilis and the utility of this intron region for phylogeny arvensis subsp, arvensis, and Scandix pecten-vene- estimation, are provided in Downie and Katz- ris) and for four enzymes (BamHI, BgIII, EcoRV, Downie (1999). All sequencing was done using an Applied Biosystem's, Inc. (Foster City, CA) 373A Automated DNA Sequencer with Stretch upgrade. The reaction conditions were as specified by the Fig. 1. Strict consensus of 588 minimal length 1,035- manufacturer, with the addition of 5% dimethyl- step trees derived from equally weighted maximum sulfoxide (DMSO). Simultaneous consideration of parsimony analysis of nuclear rDNA ITS1 and ITS2 both DNA strands across the entire region permit- sequences from 58 accessions of Heywood's (1982c) ted unambiguous base determination in nearly all Apiaceae tribe Caucalideae and related taxa using cases. 409 unambiguously aligned nucleotide positions (CIs The sequences were aligned manually and gaps with and without uninformative characters = 0.466 positioned to minimize nucleotide mismatches. and 0.428, respectively); RI = 0.756; Lee and Dow- When gap-coding was problematic, these regions nie 1999). Numbers above the nodes indicate the of the alignment were excluded from the analysis. number of times a monophyletic group occurred in Unambiguously aligned, potentially informative 100 bootstrap replicates; decay values are presented gaps were few, and were not included as extra below. Within the ingroup, three major groups of characters in the phylogenetic analysis. Pairwise taxa are recognized, and have been treated as nucleotide differences of unambiguously aligned Scandiceae subtribes Daucinae, Torilidinae, and positions were determined using the distance ma- Scandicinae (Downie et al. 2000) trix option in PAUP version 3.1.1 (Swofford 1993). 40 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae

,-.1 ,..O o ~o Ox d ~ S ~ "-" = • = ~,..., ~ ~ .~ z ,-. ~ ,-,

.,.--~ ~'" ~ ~" .~ ~ "-'~ ~, o~ ~ ~ "-" ~o,, "~ o x- ~ I-.--

o .~ r.T. ,

• ~=o ~ 0 ~'~ 0 0 0 0 0 0 0 ~,..~ 0

oO •~ ~ x~'~

,...,~ ~ .a ~ 0 0 0 0 0 0 '~ 0 • ~ .~ ~ ~.1 ~ o ~

0 0

= ~'~ .~<~o ~ ~,<.~...... "~

+._ r.,_i ,-.4 o~..~ 0 0 0 +,- =:l

o ~ ~

• .~ b,0 4~: ¢L ~ ¢L '+- ".-" . "-".~ ,,-i ~ ,,..j"~ ..~.~

o 0 II •~ ~ ~ .~ ~ ,..J '+7.

~, "~ "H "6 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae 41

o~ ~ ~ ~ ~ ~ .~ g .~ ~o .~ ~"

~.~ ~ ~- ~> ~ .~ ~.~ ~ ~ ~ ~.~ •~ ~~.~ <

¢,) 0

-d +-

8r~ =~~=+:-. ~- ~o .~~ .....~ ~~ .~~ -~~ ~ =.~o "~

~.~.~ ~ ~ ~ .= ~ ~ ~ ~.~ ~~ ~ ~ ~ ~ ~ ~ ~.~

• ~:~ "~ ~ 42 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae

i d © ,_0 z -~ ~

~.> .~-~ ~ ~ ~ Z ~.~ ~ :~

o ~

~ .~ ~~ ~ 3~.~

o m m

* 4"--- c~ ©

4~ +- • d-- 4-- 4--

-4--- 4--- ©

co B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae 43

Alignment gaps in any one sequence were treated as the chloroplast genomes of the 32 examined missing data for all taxa. Transition/transversion accessions of Apiaceae are similar in gene ratios over all maximally parsimonious trees were content, gene arrangement, and structure to calculated using MacClade version 3.01 (Maddison that of tobacco cpDNA and, thus, to the vast and Maddison 1992). The rpsl6 intron and flanking majority of angiosperms examined to date. No rpsl6 3" exon DNA sequences have been submitted length variants in the LSC and SSC regions to GenBank (accession numbers in Table 1). All were detected, but since length mutations less data matrices are available upon request; the cpDNA restriction site matrix is presented in Lee than 200 bp could not be seen easily on our gel (1998). systems, we have likely underestimated the Phylogenetie analysis. Phylogenetic analyses of actual extent of this variation. CpDNA re- the restriction site and rpsl6 intron data sets were striction site maps for four enzymes are carried out separately (as only 14 accessions were presented in Fig. 2 for Daucus carota subsp. shared between them) using the heuristic search sativus (Scandiceae subtribe Daucinae), Torilis strategies of PAUP. Twenty-five accessions were arvensis subsp, arvensis (subtribe Torilidinae), common to both the rpsl6 intron and ITS studies, and Scandix pecten-veneris (subtribe Scandic- and phylogenetic analyses of these reduced data inae). A gene map is also presented, based on sets, both separately and in combination, were also our hybridization results and those results carried out. All searches were conducted with 500 obtained using finer-scale (i.e. gene- and in- random addition replicates, tree bisection-recon- tron-specific) probes from a much larger study nection (TBR) branch swapping, with options mulpars, steepest descent, collapse, and acctran on the evolution of chloroplast genome organ- selected. Bootstrap values (Felsenstein 1985) were ization and gene content in angiosperms calculated from 100 replicate analyses, simple (Downie and Palmer 1992). Within experimen- addition sequence of taxa, and TBR branch swap- tal limits, the gene map for Daucus, Torilis, ping. In order to identify weakly supported nodes, and Scandix cpDNAs is the same as that for decay analyses (Bremer 1988) were conducted until tobacco cpDNA (Shinozaki et al. 1986, as tree storage memory was exhausted or 5,000 min- modified by Sugiura 1992 and Wolfe et al. imal length trees had been reached. We explored the 1992) and, likely, for all other examined effect of differentially weighting restriction site Apiaceae, as all probes hybridized strongly gains: losses or DNA sequence transitions: trans- and mapped in a colinear fashion. versions. The results obtained, however, did not A total of 688 restriction sites was identi- differ substantially from those under the assumption of equal weighting and are not reported further. fied using 14 enzymes: 291 (42.3%) were The maximum likelihood method was also shared by two or more taxa and were poten- applied to the 25-taxon rps16 intron and ITS data tially informative for parsimony analysis, 265 sets using the program fastDNAml (version (38.5%) were unvarying, and 132 (19.2%) were 1.0.6; Olsen et al. 1994), based on the procedures unique to individual taxa. The numbers of of Felsenstein (1981). Maximum likelihood trees parsimony informative, nonvariable, and aut- were inferred using a range of transition: transver- apomorphic restriction sites for each of the 14 sion rate ratios between 1.0 and 2.0, randomizing enzymes across all 32 cleavage maps, relative the input order of sequences (jumble), and by to their inferred positions in either the LSC or invoking the global branch swapping search op- SSC cpDNA regions, are presented in Table 2. tion. Empirical base frequencies were derived from The enzyme DraI, specific for the DNA the sequence data and used in the maximum sequence TTT'AAA, cut most frequently and likelihood calculations. yielded the greatest number of potentially informative mutations. The ratio of terminal taxa (32) to parsimony informative restriction Results sites (291) was 1:9.1. Pairwise mean distances, Chloroplast DNA restriction site analysis. The calculated using PAUP, ranged from identity results of the restriction site study reveal that (among Daucus carota subspecies halophilus, 44 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae

trnC ORF29 trnT psbD psbC ORF62 tmG psbK psbl trnG trnR IH I , • , nl , nn r--n ! II ff I s,~ll a ~ i i I i i i,,J 5,...,.i, ,, • l I B Ill LSC rpoC2 rpoC1 rpoB psbM trnDtmYtrnE trnS trnfM rps14 psaB trnH psbA tmK(matK) rps16 tmQ trnS atpA atpF atpH atpl rps2 20 25 30 35 40 o , ~ io 15 I 2 3 ~ ' 5 11 0 ~ I .... b I 10 i 11 I 12 I 13 probes I 1,2 13 ,5, 1,7 I • '91 29.599 1., 1.9 I 11.2,2 1.2 1,7 .5 • 7.7 I 5,0 3.1 BamHI 1.5 1.5 1.0 1.5 1.5 1,0 1"2 ,, 5 1,119 I 3.3 llO,9 5.0 3.1

9.0 i 1.5 3.4 ,3" -i - ' .4 2,9 3.3 1.011.0 9.0 T° 59,, il, 2, i~I 22 I 3.3 1.0 1.0 5.8 3.0 Bgm 1,4 I 2.8 2,7 6,9 4,9 3.4 .3 .... A 2.9 I 4.3 1.0 5.8 3.0 S _ 6.3 2.9 6.4 3,3 I 1,7 3.4 ,3 .... .4 2.9 11.5 .51 B DI,0 2.9 10,2 4.5 51,0 2.5 i.9 2.0 .5- 2.o 14 I 4.5 5 1.0 2.5 :.9 2.0 ,3- 2.0 14 EcoRV TI.O 2.9 3.2 I 7.0 14 S1.0 2,9 3,2 7.0 4.5 5 1.0 2,5 1,9 2.0 ~" 2.0 o11.51.51,.5 121 2.6 2,6 .3 7~ 6 11.5 4.5 ] l 7.6 6 11.5 4.5 Hindlll T 3.3 2.1 I 6.3 2,6 2.9 11.5 4.5 S 3.3 8.4 2.6 2.9 7.6 6

psal ORF229 ORF31 psaJ rps18 ORF34 petB psb8 psbH petD trnS trnL trnF trnM rbcL accD ORF184 petA petG rp133 I i ' , n r~lr'-'lr--'l nn n rlr'l , ~n.rr"~r'-"-l, I I I UULI II I.,J It l I I I I...I i r--II .... ~.,jI = i = psaA ORF168 rps4 trnT ndhJ ndhC tmV atpE atpB psbJ psbF tmW rpI20 clpP psbN rpOrApslrj~136 ndhK pSbL psbE tmP 5'rps12 75 90 45 50 55 60 65 7O 25 I 26 I 27 probes " I " I"l 1, i 19 119 J 7.1 12.0 4,3 D 5.8 ] 3.2 9.0 B 27,7 BamHI T 9,O 9.O 9,0 4,5 [ 4.5 27,7 S I ! 4.9 2.5 .3 2,0 1,7 ,4' t',31.2 1.2 48 5.4 D 2.5 4.9 2.5 ,3 2.0 1.7 .4' 1.5 1.2 .8 9,4 BgAI T 2.5 ,5 =.5.3 29 1,, M, S RK 6,6 D 7.5 6,6 EcoRV T 7,9 7,9 S 7.5 9.2 D 4,3 7.5 Hindlll T 4.3 S 4.3

LSC IR. SSC IRA rp132 tmL ORF313 II, ,. ,~ ...... 11 .....rps8 rp116 rpr~ll ndhF ndhD mINE ndhl ndhA ndhH rps15 ORF1901 InfA 11)114 rps3rps19 psaC ndhG 85 115 120 125 130 probes 2, I 2, j D 4.3 I 4.3 0 .1 6 1,2 9,4 6.6 T 11 1.= 9.4 ..9 BamHI T 27.7 S 27.7 S ~.1H 1.5 I o.9 D~,, i ~ D 1,o I ~-~ Bgld T 9,4 T 4.9 I 6.4 S 9.4 S 19.o [ s.6

D6"6 I 6,5 D 29 .S 2.3 1.8 T 6.6 6,9 T 29 161.5 22.3 i8i 1.8 EcoRV S 7.9 6.s S 2o I 8.2

i D 9.2 D s.9 [ 9.1 H/ndlll T 7.0 T s.s l 9.1 S 9"2 S 13.0

Fig. 2. Physical and gene maps of the large and small single-copy regions (LSC and SSC, respectively) of the chloroplast genomes of Daucus carota subsp, sativus (D), Torilis arvensis subsp, arvensis (T), and Scandix pecten-veneris (S). Maps of the two large inverted repeat regions, IRA and IRB, were not constructed. Chloroplast DNAs from these three taxa were each digested singly with restriction enzymes BamHI, BgIII, B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae 45 hispanicus, and maritimus, and D. maximus) to Table 2. Numbers of nonvariable (N), parsimony 22.7% (between D. pusillus and Scandix pec- informative (I), and autapomorphic (A) restriction ten-veneris), the latter representing a minimum sites derived from each of the 14 enzymes used to construct cpDNA maps for 32 accessions of of 147 site differences. Within Daucus, a Apiaceae tribe Caucalideae and related taxa. Lo- maximum of 64 restriction site differences cation abbreviations: chloroplast genome large were apparent (between D. montanus and single copy (LSC) and small single copy (SSC) D. carota subsp, sativus) for 9.3% mean diver- regions gence, and within D. carota, a maximum of seven site differences were apparent (between Restriction Number of restriction sites Total subspecies commutatus and sativus). enzyme LSC SSC Maximum parsimony analysis of the 32 taxon x 688 character matrix resulted in two N I A N I A minimal length trees each of 779 steps, with consistency indices (CIs) of 0.543 (all charac- AvaI 18 l0 8 1 2 4 43 ters) and 0.450 (excluding uninformative BamHI 8 14 4 1 0 6 33 characters) and a retention index (RI) of BanII 27 24 8 2 4 2 67 17 18 8 1 6 I 51 0.754. The strict consensus of these two trees, BglII ClaI 12 19 9 1 7 2 50 with accompanying bootstrap and decay val- DraI 16 30 10 6 9 1 72 ues, is shown in Fig. 3. Three major clades EcoOl09I 22 12 12 6 5 1 58 are inferred: Scandiceae subtribes Daucinae EcoRI 19 26 6 4 6 1 62 (Daucus, Pseudorlaya, Agrocharis, Laserpi- EcoRV 13 10 15 2 2 3 45 tium, Cuminum, and Orlaya), Torilidinae HincII 17 18 5 7 2 1 50 (Torilis, Chaetosciadium, Caucalis, Turgenia, HindIII 12 12 2 1 1 3 31 and Astrodaucus), and Scandicinae (Anthris- NciI 29 24 5 3 1 6 68 cus and Scandix). Subtribes Daucinae and XbaI 11 15 5 2 6 0 39 Torilidinae arise as strongly supported sister XhoI 6 8 1 1 0 3 19 taxa (with a 100% bootstrap value). The Total 227 240 98 38 51 34 688 genera Daucus and Torilis are each not

monophyletic: D. maximus arises within the Fig. 2 (continued) D. carota assemblage, and Chaetosciadium EcoRV, and HindIII, and probed with the 35 trichospermum occurs within Torilis. The three indicated cloned restriction fragments from tobacco examined species of Scandicinae form a strongly cpDNA. Additional information on gene arrange- supported clade sister to Daucinae + Torilidi- ment and orientation was obtained using finer-scale nae. Based on only three exemplars from probes from a much larger study on chloroplast Scandicinae, the genus Anthriscus is also not genome organization and gene and intron content in monophyletic. Bootstrap and decay values are angiosperms (Downie and Palmer 1992). Stippled generally high, ranging between 45 and 100% boxes indicate short regions not used as probes. for the former and from one to greater than four Restriction fragment sizes and coordinates above the for the latter. physical maps are in kilobase pairs (kb), with Chloroplast rpsl6 intron analysis. Among coordinates corresponding to those of the tobacco all 34 accessions examined, the rpsl6 intron chloroplast genome (Shinozaki et al. 1986). Within experimental limits, the gene map is the same as the ranged in size from 818 (Artedia) to 892 tobacco map (Shinozaki et al. 1986, as modified by (Chaetosciadium) bp, and averaged 864 bp. Sugiura 1992 and Wolfe et al. 1992) as major Percentage G + C content for the intron insertions, deletions or other structural mutations ranged from 31.4 to 34.0%, averaging 33.0%. were not evident. Genes above the line are transcribed All sequencing reactions culminated in an addi- from left to right and vice versa tional 110 bp of sequence from the adjacent 46 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae

~ Daucus carota subsp, halophilus 52 Daucus carota subsp, hispanicus Daucus carota subsp, maritimus 1O0 4~-7>9~ I Daucus maximus Daucus carota subsp, gummifer 79 Daucus carota subsp, commutatus Daucus carota subsp, sativus Pseudor[aya pumila ¢d Daucus aureus c- {- o 2 Daucus muricatus E) _~1 Daucuspusillus IO0 89 Daucus montanus >4 Agrocharis incognita Laserpitium hispidum 70 Cuminum cyminum Orlaya daucoides 100 ~ Orlaya daucorlaya >4 Orlaya grandiflora Torilis arvensis subsp, arvensis Torilis arvensis subsp, purpurea 100 ;4 Torilis leptophylla Torilis tenella >4j ¢.. Torilis nodosa .'g_°~ Chaetosciadium trichospermum o I- Torilis elongata 89 Caucalis platycarpos >4 ilOOTurgenia latifolia Astrodaucus orientalis (D 7~ Anthriscus cerefolium .~_ 97 .o_ >4 Anthriscus caucalis Scandix pecten-veneris ~o o3 Smyrniurn olusatrum

Fig. 3. Strict consensus of the two minimal length 779-step trees derived from equally weighted maximum parsimony analysis of cpDNA restriction site data (CIs = 0.543 and 0.450, with and without uninformative characters, respectively; RI = 0.754). Bootstrap percentages are provided above each branch; decay values are provided below. Three major groups of taxa are recognized, and have been designated previously as Scandiceae subtribes Daucinae, Torilidinae, and Scandicinae (Downie et al. 2000) B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae 47 rpsl6 3' exon region, with no length variation. Daucinae is representation from tribe Laser- Alignment of all 34 intron and flanking 3' exon pitieae (Melanoselinum, Monizia, Laserpitium, sequences resulted in a matrix of 1130 posi- Laser, and Polylophium). Sister to Daucinae + tions. However, due to confounding length Torilidinae is a clade comprising three acces- mutations, small repetitive elements, or tracts sions of Ferula, and is designated herein as the of poly-A's and T's of variable length, it was Ferula clade. Subtribe Scandicinae is also necessary to exclude eight regions (102 align- monophyletic, and sister to the clade compris- ment positions) from the analysis. These ing Daucinae, Torilidinae, and Ferula. The ambiguous regions ranged in size from 4 to genera Daucus (represented in this analysis by 36 positions, averaging about 13 positions D. carota subsp, sativus and D. pusillus) and each. Of the remaining 1028 unambiguously Torilis (T. arvensis subsp, arvensis and T. ja- aligned positions, 107 (10.4%) were parsimony ponica) are each, again, not monophyletic. The informative, 106 (10.3 %) were autapomorphic, two species of Laserpitium (L. hispidum and and 815 (79.3%) were invariant. The ll0-bp L. siler) also do not form a clade. 3' exon region was highly conserved, reflecting CpDNA restriction site and iTsl6 intron only five informative and three autapomorphic sequence comparisons. Comparison of phy- positions across all 34 taxa. The ratio of logenies derived from separate analysis of terminal taxa (34) to parsimony informative cpDNA restriction sites (Fig. 3) and rpsl6 nucleotide substitutions (107) was 1:3.1. Mea- intron sequences (Fig. 4) reveals much concor- sures of pairwise nucleotide sequence diver- dance of relationship, despite the fact that these gence ranged from 0.2% (between the two data sets were not parallel in construction. In accessions each of Laserpitium hispidum and each of these analyses, subtribe Daucinae is Melanoselinum decipiens) to 6.1% (between monophyletic and sister to a monophyletic Scandix and the outgroup Smyrnium). A total Torilidinae. This group, in turn, is sister to a of 37 unambiguous gaps was required for monophyletic Scandicinae when restriction proper alignment. These gaps ranged in size sites are compared, or to the Ferula clade in from 1 to 46 bp (averaging 5 bp), with the the intron-based analysis. The relative place- largest representing a deletion in Artedia rela- ments of the 14 taxa common to both analyses tive to Smyrnium. Seventeen gaps were poten- are highly consistent. As examples, Daucus tially informative for parsimony analysis, with carota subsp, sativus unites or is very closely seven of these a single bp in size. Many more allied to Pseudorlaya, D. pusillus is allied with gaps were apparent, but were in those regions Agrocharis, and Torilis and Chaetosciadium of the alignment excluded from the analysis. unite, as does Caucalis and Turgenia. Maximum parsimony analysis of the 34 Our results indicate that restriction site taxon x 1028 character matrix resulted in 52 data are more variable than those of the rpsl6 minimal length 321-step trees, with CIs of intron, even though the latter matrix includes 0.782 and 0.663 (with and without uninforma- twice as many genera. In the restriction site tive characters, respectively) and a RI of 0.779. study, 423 variable sites were scored, of which The strict consensus of these trees, with 291 were potentially parsimony-informative. accompanying bootstrap and decay values, is In the intron study, 107 of 213 variable presented in Fig. 4. Within this tree, bootstrap positions were potentially informative. The values range between 27 and 100%, and decay restriction site matrix provides almost three values are generally lower than those obtained times as many informative characters as does from restriction site data. As in the analysis of the intron data set, with the ratio of terminal restriction sites, Scandiceae subtribes Dauc- taxa to parsimony informative characters inae and Torilidinae form monophyletic sister being either 1:9.1 or 1:3.1, respectively. groups, but this relationship is poorly support- Measures of divergence, although not directly ed (with a 37% bootstrap value). Included in comparable given the different types of data 48 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae

Daucus carota subsp, sativus Pseudorlaya pumila 821 Daucus pusillus 60 ~2 Agrocharis incognita 70 Melanoselinum decipiens 1

69 Melanoselinum decipiens2 t" Monizia edulis "~ 92 Laserpitium hispidum I D 71 >2 .. Laserpitium hispidum2 1 Orlaya daucoides Laserpitium siler 100 Laser trilobum >2 Polylophium panjutinfi 981 Torilis arvensis subsp, arvensis 37._7 ~ Chaetosciadium trichospermum 1 ~l"~l~'~ I~ Torilis japonica '1 Yabea microcarpa r~ 31 1001 Turgenia latifolia .~_ "~ 69F~ Lisaea papyracea 86 27 4['~ ~1 Caucalis platycarpos 2 85 Glochidotheca foeniculacea 2 Artedia squamata 1O0 j Astrodaucus orientalis >2 ] Szovitzia callicarpa 88i -- Ferula ofivacea Ferula >2 i 98 , Ferula kokanica c ade >2 Ferula tenuisecta Anthriscus caucalis 100 >2 100 Scandix pecten-veneris "0i~ >2 Osmorhiza Iongistylis 0 Myrrhis odorata O9 Ligusticum scoticum Anisotome aromatica Smyrnium olusatrum

Fig. 4. Strict consensus of 52 minimal length 321-step trees derived from equally weighted maximum parsimony analysis of 34 unambiguously aligned rpsl6 intron and flanking 3' exon sequences (CIs = 0.782 and 0.663, with and without uninformative characters, respectively; RI = 0.779). Bootstrap percentages are provided above each branch; decay values are provided below. In addition to the three major groups of taxa (Daucinae, Torilidinae, and Scandicinae), a clade comprising three species of Ferula is evident. The two accessions each of MelanoseIinum decipiens and Laserpitium hispidum are described in Table 1 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae 49 represented, are also greater in the restriction matrices were truncated to their taxa in com- site study, with 22.7% maximum pairwise mon. Maximum parsimony analyses of these divergence vs. 6.1% when the intron sequences separate, reduced data sets resulted in a single are compared. As a consequence of the greater shortest tree for the ITS matrix (Fig. 5A; tree number of potentially informative mutations, length = 643 steps; CIs = 0.563 and 0.478, the branches in the strict consensus tree with and without uninformative characters; derived from restriction site data (Fig. 3) are RI = 0.617) and 50 minimal length trees for generally better supported (with higher boot- the intron matrix (tree length = 275 steps; strap and decay values) than those of the rpsl 6 CIs = 0.807 and 0.669, with and without intron tree (Fig. 4). For example, in the uninformative characters; RI = 0.758); the restriction site tree, the clade of Daucinae + latter were used to construct a strict consensus Torilidinae is supported with a 100% boot- tree (Fig. 5B). Once more, the position of strap value and a decay index > 4; in the intron Artedia varies depending upon DNA region tree, this clade is supported poorly, with a analyzed, with affinities to both Daucinae and bootstrap value of 37% and a decay index of Torilidinae apparent. All other areas of dis- one. Both data sets reveal comparable levels of cord occur within subtribe Torilidinae, and homoplasy, as assessed by similar RI values involve the relative positions of Glochidotheca, (0.754 or 0.779). Astrodaucus, and Szovitsia. In the intron tree Chloroplast DNA and ITS sequence com- (Fig. 5B), GIochidotheca is sister to the clade of parisons. A study incorporating data from the Turgenia, Lisaea, and Caucalis, whereas in the nuclear rDNA ITS region was carried out ITS tree (Fig. 5A), Glochidotheca allies with prior to the two plastid DNA studies presented Astrodaucus and Szovitsia. The basal position herein, and the reader is referred to Lee and of Astrodaucus and Szovitsia in subtribe Tor- Downie (1999) for details of the phylogenetic ilidinae, as inferred by the intron study, is not analyses conducted (the results of the maxi- evident in the ITS tree. mum parsimony analysis are presented in In both the ITS- and rpsl6 intron-derived Fig. 1), evolutionary characteristics of the trees (Fig. 5A and 5B, respectively), branches ITS sequences, and accession voucher infor- within the Torilidinae clade, with the exception mation. With the exception of Artedia, which of those leading to the clade of Turgenia, is sister to the Daucinae clade in the ITS tree Lisaea, and Caucalis (with 74 or 91% boot- (albeit with very weak bootstrap support; strap values) or to the clade of Torilis arvensis Fig. 1) or is nested within the Torilidinae clade and Chaetosciadium (with 100% bootstrap in the rpsl6 intron tree (Fig. 4), phylogenetic values), are weakly supported, with bootstrap analyses of each of the three data sets (ITS, values ranging between 28 and 66%. When rpsl6 intron, and restriction sites) reveal three those nodes characterized by bootstrap values major clades of similar composition (i.e. <66% and a decay index of one are treated as Daucinae, Torilidinae, and Scandicinae). unresolved (that is, they are collapsed to yield However, the sister group relationship between polytomies), the relationships among members subtribes Daucinae and Torilidinae, evident in of the Torilidinae clade are fully consistent the analyses of both plastid DNA data sets (albeit largely unresolved), with Artedia being (Figs. 3-4), is not apparent in the ITS tree the only remaining element of discord. The (Fig. 1). general agreement between the results of these Twenty-five accessions were common to separate analyses suggested that a combined both the ITS and rpsl6 intron data sets (as analysis would likely lead to the best estimate opposed to the 13 accessions shared among all of phylogeny given the available data (Barrett three data sets), so to more readily compare et al. 1991, Bull et al. 1993), although we are the phylogenetic relationships as inferred from well aware of the controversy surrounding this these nuclear- or plastid-derived data, the issue (reviewed by De Queiroz et al. 1995, 50 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae

97r~ Daucus carota sativus Daucus carota sativus 91 I :>°~'" Pseudorlaya pumila __ 52 1>L , Pseudorlayapumila A ~? Daucuspusillus B ~Daucuspusfl/us 501 >5 I ~ Agrocharisincognita cO I 2 I >~ ~ Agrocharis incognita c c r~ ~ Laserpitium hispidum ' Laserpitium hispidum "I N 216 Orlaya daucoides cO ~ Orlaya daucoides o~ £3 £3 I ' ~/-- Laserpitium sfler ] ~ ~ - - Laserpitium siler 63._J 1L194r.._Laser trilobum I I oo i Lasertrilobum 11 ~L._ Polylophium panjutinii I >2 I Polylophium panjutinfi I ..... Arteclia squamata I 100.. I' Torilisarvensis = lOr._~ Torilis arvensis 7>--~2 >2 I Chaetosc.trichospermum 3 L-- Chaetosciadium trichospermum Yabea microcarpa ~ ~ Yabeamicrocarpa Turgenia latifolia r" 32 , 8~66 GIochidotheca foeniculacea .-- Lisaea papyracea

Astrodaucus orientalis r,z_.r Caucalis platycarpos .t ' Szovitsia callicarpa o 81 Glochidotheca foeniculacea o F- "3" 8~ lO0.~.O~ Turgenia latifolia Artedia squamata 9_]] >5 L.-- Lisaea papyracea 1 Ioc100 r-- Astrodaucus orientalis 5 L Caucalis platycarpos >2 i Szovitsia callicarpa -- Q 5ooF~ Anthriscus caucafis --7 7r~1[-- Myrrhis odorata "~ 99 Mfrrhis odorata .~ >2 Osmorhiza Iongistylis "~c ' • Scandix pecten-veneris ~o Scandix pecten-veneris Ligusticum scoticum O0 Ligusticum scoticum or) F Smymium olusatrum Smyrnium olusatrum F

51' 97 I~ >u u_ PseudorlayaDaucus carota pumila sativus ~p 100 Oaucuspusillus 66 ] >9 ] ~ ' Agrocharis incognita t- Fig. 5. Cladograms resulting from separate and N ~-- Laserpitium hispidum "~ ~ Orlaya daucoides combined maximum parsimony analyses of nuclear 4~Jr'---- Laserpitium slier rDNA ITS and cpDNA rps16 intron and flanking "141002]~ Laser.... trilobum 3' exon sequences for 25 taxa of Caucalideae and >9 L._ Polylophium panjutinii relatives. Bootstrap support is shown above branches, ~ Torilis arvensis decay values below. (A) The single maximally parsimonious tree of 643 steps resulting from separate 91 Chaetosc. trichospermum -g Yabea microcarpa analysis of ITS sequence data; CIs of 0.563 and 0.478, 7~386~Glochidotheca foeniculacea with and without uninformative characters; RI of Astrodaucus orientalis .c_ 0.617. (B) Strict consensus of 50 minimal length 275 92 "1o 6 .... Szovitsia calficarpa step trees resulting from analysis of rps16 intron and ~ rurgenia latifolia #_ flanking 3' exon sequences; CIs of 0.807 and 0.669, 97 7 Lisaea papyracea with and without uninformative characters; RI of 9 L_ Caucalis,., ..... platycarpos 0.758. (C) The single maximally parsimonious tree of Artedia squamata 923 steps resulting from combined analysis of ITS 37~-'-- Anthriscuscaucalis 7 and rpsl6 intron data; CIs of 0.633 and 0.518, with ~ IL-- Myrrhis odorata "~ and without uninformative characters; RI of 0.644. 100 [- ~ Osrnorhiza Iongistylis :~c Complete taxon names are provided in Table 1; >9 Scandix pecten-veneris ~0 subtribal circumscription based on Downie et al. Ligusticum scoticum (2000). The variably positioned Artedia squamata is V Smymium olusatrum boldfaced B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae 51

Huelsenbeck et al. 1996, Wendel and Doyle intron varies considerably in length, ranging 1998). Parsimony analysis of the combined ITS between 707 and 951 bp in size (Oxelman et al. and intron data sets resulted in a single most 1997), and our report of 818 to 892 bp for this parsimonious tree of 923 steps (CIs = 0.633 region is consistent with these values. The and 0.518, with and without uninformative intron's average G + C content of 33.0% is characters; RI = 0.644). The topology of this similar to that reported for other umbellifer tree (Fig. 5C) is very similar to that inferred rpsl6 introns (Downie and Katz-Downie using ITS data alone, with the following 1999), and falls near the GC range reported exceptions: (1) Artedia is now sister to the for vascular plant chloroplast genomes in Torilidinae clade; and (2) Astrodaucus and general (Palmer 1991). Like other group II Szovitsia are monophyletic and sister to Glo- introns, the rpsl6 intron is characterized by six chidotheca. For those clades identified in the major structural domains (Michel et al. 1989). separate analyses that were not in conflict, Domains V and VI and portions of domain I bootstrap and decay values for these clades are are necessary for proper intron processing and generally higher upon consideration of all are most conserved evolutionarily, whereas available data. To better understand what domains II, III, and IV can be quite variable effect Artedia had upon the resultant phyloge- (Learn et al. 1992, Downie et al. 1998). In this netic hypothesis, this taxon was removed and study, the smallest rpsl6 intron occurs in the parsimony analysis of combined data Artedia and reflects the removal of approxi- rerun. Here, two maximally parsimonious trees mately half of domain IV. With the exceptions resulted (tree length = 861 steps; CIs = 0.643 of domains V and VI, where no length and 0.530, with and without uninformative mutation and a high degree of sequence characters; RI = 0.664) and their strict con- conservation are apparent, all unambiguous sensus, with the exception of the collapse of the alignment gaps and regions excluded from the branch leading to Laserpitium siler, was iden- analyses were distributed equally throughout tical to the tree inferred when Artedia was the intron. included (figure not shown). Chloroplast genome evolution. Previous Long-branch attraction in parsimony anal- studies have determined that the sizes of ysis may affect the resulting phylogenetic Apiaceae chloroplast genomes range between hypothesis. Given that the branch leading to 140 and 155 kb, the largest being reported for Artedia is one of the longest in the trees, Daucus (Debonte et al. 1984, Plunkett and maximum likelihood analyses of separate and Downie 1999). This variation is largely attrib- combined data sets were carried out. However, utable to the expansion or contraction of the with respect to the phylogenetic placement of inverted repeat into the adjacent LSC region Artedia, the results obtained using maximum (Plunkett and Downie 1999). Because we did likelihood (not shown) were identical to those not survey for restriction site variation in the inferred by maximum parsimony. Once more, inverted repeat, estimates of genome size for Artedia arises sister to Daucinae in the ITS tree, Scandiceae are not available. Nevertheless, as sister to Torilidinae in the combined tree, or major insertions, deletions, or other structural included within Torilidinae in the intron tree. mutations were not evident in both LSC and SSC regions, the sizes of these single-copy regions are comparable to those reported for Discussion tobacco and many other angiosperm chloro- Molecular characteristics of the rpsl6 in- plast genomes (Shinozaki et al. 1986, Downie tron. The chloroplast gene rpsl6, encoding and Palmer 1992). Within experimental limits, ribosomal protein S16 (Neuhaus et al. 1989), the gene map proposed in Fig. 2 for Daucus is interrupted by an intron in many different carota subsp, sativus (the common cultivated land plants (Downie and Palmer 1992). This carrot), Torilis arvensis subsp, arvensis, and 52 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae

Scandix pecten-veneris is also the same as that we have found that restriction site analysis of the tobacco chloroplast genome, as all provides 3-4 times more variable characters probes hybridized in a colinear manner (Shino- than does DNA sequencing across a compa- zaki et al. 1986, as modified by Sugiura 1992 rable array of taxa. Similar results have been and Wolfe et al. 1992). The lack of major obtained by Kim et al. (1992) for Asteraceae. deletions in the restriction site maps of all As a consequence, the trees inferred herein other examined Apiaceae relative to these using restriction site data are generally more three taxa and tobacco indicates that they all resolved and the clades better supported than possess the same complement of genes and those derived from DNA sequences, with both introns in the same order. data sets revealing comparable levels of ho- Utility of epDNA restriction site data. moplasy across the shortest trees inferred. Phylogenetic analysis of separate cpDNA Moreover, restriction site data appear to be restriction site and rpsl6 intron sequence data better suited for resolving relationships at sets reveals relationships that are largely infrageneric levels within Apiaceae (such as consistent among the 14 accessions common those within Daucus and Torilis) than any to both studies. Similarly, a concurrent but existing DNA sequence data set. As a conse- higher-level study of cpDNA restriction site quence of the greater utility of the restriction variation (Plunkett and Downie 1999), includ- site data, we are continuing to use these data to ing representation from all three subfamilies of further resolve relationships within Daucus Apiaceae and related Araliaceae, yields trees (Lee and Downie, unpubl, data). that are congruent to those derived from the Summary of relationships. Based on phylo- comparative analyses of chloroplast matK or genetic analyses of molecular data, Apiaceae rpoC1 intron sequences across a similar group tribe Caucalideae (Heywood 1982c) comprises of taxa (Plunkett et al. 1996, Downie et al. two major lineages, designated as Scandiceae 1998). While this congruence of relationship subtribes Daucinae and Torilidinae (Downie supports the robustness of the phylogenetic et al. 2000). Variously associated with these hypotheses inferred, it is realized that because two subtribes are genera belonging to Hey- these restriction site and sequence data are wood's (1971b) Scandiceae (our Scandiceae derived from the same chromosome they are subtribe Scandicinae), and while evidence from inherited as a single linkage group and, hence, cpDNA points unequivocally to a sister group should have the same evolutionary history relationship between Scandicinae and the clade (Doyle 1992). Incongruences among the trees of Daucinae + Torilidinae, the ITS data do inferred, if they exist, may be due to several not. Therefore, the recognition of three distinct factors, such as differences in sampling or yet closely related groups, as proposed in an heterogeneity of evolutionary rates. earlier investigation (Downie et al. 2000), is Phylogenetic studies incorporating DNA maintained. Because the results of our previ- sequences have largely replaced those based on ous ITS study were largely concordant to those comparative restriction site mapping, largely a inferred herein on the basis of cpDNA evi- result of the ease of obtaining sequence data dence, a detailed discussion of phylogenetic both through automated sequencing methods relationships has already been presented (Lee and PCR technology. In spite of the problems and Downie 1999). Further discussion will be inherent in using cpDNA restriction site data limited to those taxa and relationships unique for phylogenetic analysis (reviewed in Olin- to this study. stead and Palmer 1994 and Mishler et al. Laserpitieae. The incorporation of mem- 1996), the method has potential to yield more bers of Drude's (1897-1898) Laserpitieae (i.e. phylogenetically informative characters than Laser, Laserpilium, Melanoselinum, Monizia, does comparative DNA sequencing. In this and Polylophium) into Scandiceae subtribe study, as that of Plunkett and Downie (1999), Daucinae is consistent, in part, with the B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae 53 classification systems of Calestani (1905) and Comparison to Drude's treatment. Drude Koso-Poljansky (1916). These systems, relying (1897-1898) defined Scandiceae on the basis of almost exclusively on anatomical characters of calcium oxalate (druse) crystals in the paren- the mericarp, indicated a close relationship chyma cells surrounding the carpophore and between Laserpitium and Daucus (and other divided it into two subtribes, Caucalidinae and taxa), treating them together in tribe Dauceae Scandicinae, according to the shape of the or subtribe Daucinae. Additional evidence fruit. Both tribe Dauceae and subtribe Cauca- suggesting that members of Daucinae and lidinae were characterized by spinose fruit Laserpitieae are closely related include the ridges, with the former allied to tribe Laser- presence of similar morphological characters, pitieae, whose members have fruits without such as distinctive secondary ridges and dor- spines but with primary and prominent sec- sally compressed fruits with single rows of ondary ridges (that are often extended into appendages (Tamamschjan 1947). Given that wings). The secondary fruit ridges of many all examined members of Drude's (1897-1898) Caucalidinae are suppressed or less well devel- tribe Laserpitieae fall within subtribe Dauc- oped than those of tribe Dauceae; members of inae, its few remaining members (Distichoseli- Scandicinae lack both secondary ridges and hum, Elaeoselinum, GuiIlonea, Margotia, spines. Drude assumed that the secondary Rouya, Thapsia, and Tornabenea; Pimenov spinose ridges in his presumably divergent and Leonov 1993) are deservant of further Caucalidinae and Dauceae had evolved inde- study. Based on our results (e.g. Fig. 1), how- pendently. Our results indicate clearly that ever, it is unlikely that Laserpitieae constitutes Drude's Caucalidinae and Dauceae should be a monophyletic group within Daucinae. united, and that these spiny-fruited umbellifers Scandicinae. Our previous study of Cau- constitute a closely related group, a relation- calideae ITS sequences (Lee and Downie 1999) ship in accordance with the earlier classifica- included five genera in Scandiceae subtribe tion systems of Bentham (1867) and Boissier Scandicinae (Anthriscus, Kozlovia, Osmorhiza, (1872) and the more recent treatment of Scandix, and Myrrhis); fewer genera were Heywood (1982c). Our results indicate further included in each of the two plastid DNA that Drude's spineless Laserpitieae allies studies presented herein. While monophyly of strongly with our subtribe Daucinae, indicat- Scandicinae is supported strongly in all ing that the presence of both prominent analyses, with bootstrap values of 97-100%, secondary ridges and strongly dorsally com- the generic level relationships within this clade pressed fruits are important synapomorphies differ (likely due to sample size and different defining this clade. Considering Fig. 5C, mer- genera sampled). For example, results of the icarp spines are inferred to have been lost at rps16 intron analysis (Fig. 4) place Scandix least twice independently in Daucinae (i.e. in sister to Osmorhiza, whereas the ITS study Laserpitium hispidum and the clade of L. siler, (Fig. 1) places Scandix sister to the clade of Laser trilobum, and Polylophium panjutinii). Anthriscus, Myrrhis, Kozlovia, and Osmorhiza. Additional discussion on the evolution of Moreover, the restriction site analysis (Fig. 3) mericarp morphological and anatomical fea- reveals that Anthriscus is not monophyletic. tures in light of the group's inferred evolu- Expanded sampling, incorporating ITS se- tionary history is forthcoming (Lee et al., ms. quence data from all 18 genera (82 accessions) in prep.). commonly treated in tribe Scandiceae, not Artedia. The most striking difference be- only supports monophyly of Scandicinae (up- tween the plastid- and ITS-derived trees is the on the exclusion of Grammosciadium and placement of the monotypic Artedia, with Rhabdosciadium), but also reveals that Anth- affinities to both Daucinae and Torilidinae riscus is indeed monophyletic (Downie et al. apparent. This discordance may be the result 2000). of several factors, such as the effects of lineage 54 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae sorting or convergence, different modes of Ferula. The placement of three species of inheritance, or chloroplast capture through Ferula alongside Daucinae and Torilidinae - a hybridization-introgression events (Doyle relationship proposed initially on the basis of 1992, Rieseberg and Soltis 1991). At this point neighbour-joining analysis of ITS data (Vali- in time we cannot offer an explanation for this ejo-Roman et al. 1998), confirmed through phylogenetic incongruence or to state which independent ITS sequencing (Downie, unpubl. phylogeny, if any, most accurately reflects data), and now supported using rpsl6 intron organismal relationships. With very few re- data - is intriguing. Ferula is a large, morpho- ported cases, interspecific hybridization hasn't logically variable genus of some 170 species been considered an important factor among (Pimenov and Leonov 1993), considered allied extant umbellifers (Heywood 1982b), but this to Peucedanum (Pimenov 1982). Indeed, our may be simply due to the lack of studies previous molecular phylogenetic investigations specifically set up to identify such an event. place F. assa-foetida L. and F. communis L. With its lateral secondary ridges developed alongside Peucedanum and related taxa, well into deeply lobed, scaly, expanded wings, away from tribe Scandiceae (Downie et al. Artedia is morphologically anomalous in the 1998, Katz-Downie et al. 1999). The mono- group. The fruit of Artedia has been consid- phyly of Ferula, supported most recently by ered highly specialized evolutionarily (A1-Attar Shneyer et al. (1995), is now brought into 1974), yet this specialization is not reflected in question. We are at a loss to explain the many of its other characters, whether these be association between the three Ferula taxa from flavonoids, volatile oils, stomates, seed- included in this study and Scandiceae. No lings, or pollen grains (Cerceau-Larrival 1962, obvious morphological or anatomical charac- Crowden et al. 1969, Harborne and Williams ters support this relationship, nor has such an 1972, Williams and Harborne 1972, Guyot association been considered in any taxonomic et al. 1980). Additional studies are in order. work to date of which we are aware. Further- Glochidotheca, Astrodaucus, and Szovitsia. more, both nuclear- and chloroplast-derived The alliance among Glochidotheca [syn. Tur- data point to the same relationship. Additional geniopsis foeniculacea (Fenzl) Boiss.], Astro- studies are necessary before we can accept the daucus, and Szovitsia, inferred on the basis of inclusion of Ferula within Scandiceae. separate analysis of ITS sequences (Figs. f and Chaetosciadium. In each of our analyses, 5A) or combined analysis of ITS and rpsl6 the monophyly of Torilis is strongly supported intron data (Fig. 5C), is surprising given the if its boundary is expanded to include the remarkable differences seen in their mericarp monotypic Chaetosciadium. Chaetosciadium is anatomy and in the greatly different shapes of characterized by mericarps that are irregularly their secondary appendages. We have covered with fine, long bristly hairs and with observed, however, that this group can be obsolete secondary ridges. These unique fea- characterized by two nonmolecular synapo- tures led Calestani (1905) to erect the mono- morphies: the presence of curved primary hairs typic subtribe Chaetosciadieae of tribe and the presence of peg-like projections on the Ligusticeae. Chaetosciadium and Torilis, how- surface of their secondary appendages. In all ever, share similar flavone distribution pat- trees presented, resolution within the Torili- terns (Crowden et al. 1969, Harborne and dinae clade is poor, with many weakly sup- Williams 1972) and hairs on their primary ported nodes. Additional molecular data are fruit ridges (Heywood and Dakshini 1971). necessary to confirm both the monophyly of They also share a base chromosome number of Glochidotheca, Astrodaucus, and Szovitsia (as six, a number rare in Apiaceae subfamily suggested by these morphological characters) Apioideae where x = 11 prevails (Constance and to better assess their placement within the et al. 1971, Moore 1971). McNeill et al. (1969) subtribe. revealed the strong similarity between Chae- B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae 55 tosciadium and Torilis using numerical analys- its subspecific status be resumed. If this is es of primarily fruit, leaf, and inflorescence done, D. carota is monophyletic. On the basis characters. Zohary (1972) was of the opinion of morphological and chemical data, Pseudor- that Chaetosciadium should probably be in- Iaya purnila is very similar to Daucus (Har- cluded within the genus Torilis, and on the borne et al. 1969, Heywood and Dakshini basis of our molecular results we concur. A 1971, Williams and Harborne 1972, Lee et al., name for such a combination has already been unpubl, data), providing additional evidence published, Torilis trichosperrna (L.) Spreng. for the transfer of P. purnila into Daucus. We (Umb. Spec. 142). wish, however, to examine the two remaining Daucus. Within Daucus the number of species of Pseudorlaya and additional material species varies considerably, with 21 species in of P. pumila before nomenclatural changes are seven sections recognized by Heywood made. The position of the monotypic Pachy- (1982c). A similar situation occurs within the ctenium has yet to be considered using cpDNA D. carota complex, with the variously delim- data, although on the basis of ITS sequences it ited subspecies extremely difficult to circum- is also clearly positioned within Daucus. The scribe unambiguously (Heywood 1968b, Small ITS phylogeny (Fig. 1) indicates a major 1978, St. Pierre et al. 1990). In our ITS study dichotomy within Daucus, with D. bicolor, (Lee and Downie 1999; Fig. 1), nine species D. pusillus, D. montanus, and D. durieua from all of Heywood's seven sections were forming one clade and all remaining examined represented, including four subspecies of Daucus species the other; a similar relationship D. carota and two subspecies of D. bicolor. is seen in the restriction site tree (Fig. 3), but In the cpDNA restriction site study (Fig. 3), with fewer taxa represented. The close six species were considered, including six relationship between Agrocharis, the only subspecies of D. carota. The results of these genus of Caucalideae endemic to tropical two studies were similar in showing: (1) a close Africa (Heywood 1982c), and Daucus reflects relationship between the only two New World the similarities observed in their fruit anatomy species within the genus, D. pusitlus and and morphology (Jury 1986, Lee etal., D. montanus; (2) an affinity between these unpubl, data). Indeed, Agrocharis was first New World taxa and the eastern tropical described as a species of Daucus (reviewed in African genus Agrocharis; (3) close relation- Heywood 1973) and its return to the latter ships between D. carota and D. rnaximus, and needs to be investigated further. Clearly, both between D. aureus and D. muricatus; and (4) ITS sequencing and the mapping of cpDNA the inclusion of Pseudorlaya purnila within the restriction sites are useful in resolving inter- Daucus clade. In our ITS study, Pachyctenium specific relationships within Daucus; additional mirabile arises within the Daucus group, phylogenetic studies are currently being carried D. crinitis is allied closely with D. aureus and out in order to address the aforementioned D. muricatus, and D. bicolor, D. pusillus, taxonomic issues. D. rnontanus, and D. durieua form a strongly supported clade. D. carota subsp, sativus, the Conclusions domestic garden carrot, is allied strongly with D. carota subsp, carota. Phylogenetic analyses of cpDNA restriction Daucus rnaxirnus, recognized initially as a site polymorphisms and rps 16 intron sequences subspecies of D. carota (D. carota subsp. from representatives of Heywood's (1982c) maximus (Desf.) Ball), was treated as a distinct tribe Caucalideae and related taxa yield trees species by Heywood and Saenz de Rivas (1974) largely congruent to each other and to hy- and Heywood (1982c). Our restriction site potheses of relationship based on nuclear study (Fig. 3) clearly positions this taxon rDNA ITS sequences (Lee and Downie within D. carota and, as such, we suggest that 1999). Three major lineages of equivocal 56 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae relationship are inferred, coinciding with the tionships for those taxa in common, greater previously delimited Scandiceae Spreng. sub- resolution and higher branch support was tribes Daucinae Dumort., Torilidinae Du- achieved using restriction site data. Additional mort., and Scandicinae Tausch (Downie et al. mapping studies, incorporating data from both 2000). Subtribes Daucinae and Torilidinae chloroplast and nuclear genomes, and ITS coincide with Heywood's circumscription of sequencing are currently being pursued in Caucalideae, but with the inclusion of Drude's order to clarify relationships within the poly- (1897-1898) tribe Laserpitieae in the former morphic Daucus and their close relatives, such and the transfer of Kozlovia to Scandicinae. as Pseudorlaya, Pachyctenium, and Agrocharis. Subtribe Scandicinae coincides roughly with Heywood's (1971 b) circumscription of Scandi- The authors thank M. Chase, E. Knox, J. ceae; further details of relationships within Lahham, M. Pimenov, J.-P. Reduron, and M. Scandicinae are presented in Downie et al. Watson, and the many botanical gardens cited in the text for generously providing us with leaf, seed (2000). Aphanopleura and Psammogeton, treat- or DNA material; K-J. Cho, M. Choi, and D. ed previously in Caucalideae (Heywood Katz-Downie for assistance in the laboratory; and 1982c), are distantly related to the group D. Katz-Downie, S. Jury, and one anonymous (Katz-Downie et al. 1999). The only major reviewer for comments on the manuscript. This difference between the ITS- and chloroplast- work was supported by NSF grant DEB-9407712 derived phylogenies is the position of the to SRD and by a H. H. Ross Award to BYL. This morphologically anomalous Artedia, affiliated paper represents a portion of a Ph.D, dissertation weakly with either the Daucinae or Torilidinae submitted by BYL to the Graduate College of the clades. Additional data are necessary to re- University of Illinois at Urbana-Champaign. solve the phylogenetic placement of this genus, as well as to clarify relationships within References subtribe Torilidinae, particularly among the genera Astrodaucus, Glochidotheca, and Szovit- A1-Attar A. (1974) Studies in the systematic anat- sia. A most unusual and unexplainable find is omy, embryology and morphology of the Um- the close relationship among three species of belliferae tribe Caucalideae. Ph.D. thesis, Ferula and Scandiceae in the rpsl6 intron- University of Reading: Reading. Barrett M., Donoghue M. J., Sober E. (1991) based analysis, a relationship reported else- Against consensus. Syst. Zool. 40: 486M93. where on the basis of nuclear rDNA ITS data Bentham G. (1867) Umbelliferae. In: Bentham G., (Valiejo-Roman et al. 1998). Extended molec- Hooker J. D. (eds.) Genera Plantarum. Reeve, ular, morphological, and cytological studies London 1: 859-931. are underway to resolve and explain this Boissier E. (1872) Umbelliferae. In: Flora oriental- intriguing relationship. Detailed morphologi- is. Georg, Gen6ve 2: 819-1090. cal and anatomical investigations are also Bremer K. (1988) The limits of amino acid underway, and when completed will provide sequence data in angiosperm phylogenetic re- insight into character evolution, including the construction. Evolution 42: 795-803. identification of morphological synapomor- Bull J. J., Huelsenbeck J. P., Cunningham C. W., phies supporting each of the major clades Swofford D. L., Waddell P. J. (1993) Partition- identified herein on the basis of molecular ing and combining data in phylogenetic analysis. Syst. Biol. 42: 384-397. data. Calestani V. (1905) Contributo alla sistematica This study demonstrates further the phylo- delle Ombrellifere d'Europa. Webbia 1: 89-280. genetic utility of chloroplast rpsl6 intron Cauwet-Marc A.-M., Carbonnier J. (1982) Les sequences, a region which has seen very little Ombellif6res. Actes du 2 +me Symposium Inter- use in molecular phylogenetic studies to date. national sur les Ombellif6res "Contributions However, while both cpDNA restriction site Pluridisciplinaires ~t la Syst6matique." Mono- and intron data sets suggested similar rela- graphs in Systematic Botany from the Missouri B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae 57

Botanical Garden, vol. 6. Braun-Brumfield, Ann aceae subfamily Apioideae: phylogenetic analy- Arbor. sis of nuclear ribosomal DNA internal Cerceau-Larrival M.-Th. (1962) Plantules et pol- transcribed spacer and plastid rpoC1 intron lens d'Ombellif6res. Leur inter& syst~matique et sequences. Amer. J. Bot. 85: 563-591. phylog6nique. Th~se. M6m. Mus. Natl. Hist. Doyle J. J. (1992) Gene trees and species trees: Nat. S6r. B (Bot,) 14: 1-166. molecular systematics as one-character taxono- Cerceau-Larrival M.-Th. (1965) Le pollen d'Om- my. Syst. Bot. 17: 144-163. bellif6res M6diterran&nnes. Ill. Scandicinae Doyle J. J., Doyle J. L. (1987) A rapid DNA Drude. IV. Dauceae Drude. Pollen et Spores 7: isolation procedure for small quantities of fresh 35-62. leaf tissue. Phytochem. Bull. 19: 11-15. Constance L., Chuang T.-I., Bell C. R. (1971) Drude O. (1897-1898) Umbelliferae. In: Engler A., Chromosome numbers in Umbelliferae. IV. Prantl K. (eds.) Die natfirlichen Pflanzenfamilien Amer. J. Bot. 58: 577-587. Wilhelm Engelmann, Leipzig 3: 63-250. Crowden R. K., Harborne J. B., Heywood V. H. Felsenstein J. (1981) Evolutionary trees from DNA (1969) Chemosystematics of the Umbelliferae - a sequences: a maximum likelihood approach. general survey. Phytochem. 8: 1963-1984. J. Mol. Evol. 17:368 376. Debonte L. R., Matthews B. F., Wilson K. G. Felsenstein J. (1985) Confidence limits on phylog- (1984) Variation of plastid and mitochondrial enies: an approach using the bootstrap. Evolu- DNAs in the genus Daucus. Amer. J. Bot. 71: tion 39: 783-791. 932-940. Guyot M., Cerceau-Larrival M.-T., Carbonnier- De Queiroz A., Donoghue M. J., Kim J. (1995) Jarreau M.-C., Derouet L., Relot J. (1980) Separate versus combined analysis of phyloge- Correlations entre types stomatiques et types netic evidence. Annu. Rev. Ecol. Syst. 26: 657- polliniques dans la tribu des Caucalid&s (Om- 681. bellif~res). Bull. Mus. Natn. Hist. Nat., Paris, Downie S. R., Palmer J. D. (1992) Use of chloro- sect. B 4: 341-385. plast DNA rearrangements in reconstructing Harborne J. B., Williams C. A. (1972) Flavonoid plant phylogeny. In: Soltis P. S., Soltis D. E., patterns in the fruits of the Umbelliferae. Phy- Doyle J. J. (eds.) Molecular systematics of plants. tochem. 11: 1741-1750. Chapman and Hall, New York, pp. 14-35. Harborne J. B., Heywood V. H., Williams C. A. Downie S. R., Katz-Downie D. S. (1996) A (1969) Distribution of myristicin in seeds of the molecular phylogeny of Apiaceae subfamily Umbelliferae. Phytochem. 8: 1729-1732. Apioideae: evidence from nuclear ribosomal Heywood V. H. (1968a) Scanning electron micros- DNA internal transcribed spacer sequences. copy and micro-characters in the fruits of the Amer. J. Bot. 83: 234-251. Umbelliferae-Caucalideae. Proc. Linn. Soc. Downie S. R., Katz-Downie D. S. (1999) Phyloge- Lond. 179: 287-289. netic analysis of chloroplast rpsl6 intron se- Heywood V. H. (1968b) Daucus. In: Tutin T. G. quences reveals relationships within the woody et al. (eds.) Flora Europaea, vol. 2. University southern African Apiaceae subfamily Apioideae. Press, Cambridge, pp. 373-375. Can. J. Bot. 77: 1120-1135. Heywood V. H. (1971a) The Biology and Chemis- Downie S. R., Katz-Downie D. S., Cho K.-J. try of the Umbelliferae, Supplement 1 to the (1996) Phylogenetic analysis of Apiaceae sub- Botanical Journal of the Linnean Society. Aca- family Apioideae using nucleotide sequences demic Press, New York. from the chloroplast rpoC1 intron. Molec. Phy- Heywood V. H. (1971b) Systematic survey of Old lo. Evol. 6: 1-18. World Umbelliferae. In: Heywood V. H. (ed.) Downie S. R., Katz-Downie D. S., Spalik K. (2000) The Biology and Chemistry of the Umbelliferae, A phylogeny of Apiaceae tribe Scandiceae: Supplement 1 to the Botanical Journal of the evidence from nuclear ribosomal DNA internal Linnean Society, vol. 64. Academic Press, New transcribed spacer sequences. Amer. J. Bot. 87: York, 31-41. 76-95. Heywood V. H. (1971c) Chemosystematic studies Downie S. R., Ramanath S., Katz-Downie D. S., in Daucus and allied genera. Boissiera 19: 289- Llanas E. (1998) Molecular systematics of Api- 295. 58 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae

Heywood V. H. (1973) The taxonomic position of phylogeny of Apiaceae subfamily Apioideae: Agrocharis Hochst. and allied genera. Notes additional information from nuclear ribosomal Roy. Bot. Gard. Edinburgh 32: 211-215. DNA ITS sequences. Plant Syst. Evol. 216: 167- Heywood V. H. (1982a) History and aims of the 195. symposium. In: Cauwet-Marc A.-M., Carbon- Kim K. J., Jansen R. K., Wallace R. S., Michaels nier J. (eds.) Les Ombellif~res. Actes du 2 +me H. J., Palmer J. D. (1992) Phylogenetic implica- Symposium International sur les Ombellif6res tions of rbcL sequence variation in the Astera- "Contributions Pluridisciplinaires fi la Syst6ma- ceae. Ann. Missouri Bot. Gard. 79: 428-445. tique." Monographs in Systematic Botany from Koso-Poljansky B. M. (1916) Sciadophytorum the Missouri Botanical Garden, vol. 6. Braun- systematis lineamenta. Bull. Soc. Imp. Nat. Brumfield, Ann Arbor. 1-4. Moscou 29: 93-222. Heywood V. H. (1982b) General introduction to Koso-Poljansky B. M. (1917) Sciadophytorum the taxonomy of the Umbelliferae. In: Cauwet- systematis lineamenta. Mantissa prior. Bull. Marc A.-M., Carbonnier J. (eds.) Les Ombel- Soc. Imp. Nat. Moscou 30: 277-290. lif~res. Actes du 2 ~rne Symposium International Learn Jr. G. H., Shore J. S., Furnier G. R., sur les Ombellif@es "Contributions Pluridisci- Zurawski G., Clegg M. T. (1992) Constraints on plinaires /t la Syst6matique." Monographs in the evolution of plastid introns: the group II Systematic Botany from the Missouri Botanical intron in the gene encoding tRNA-Val(UAC). Garden, vol. 6. Braun-Brumfield, Ann Arbor. Mol. Biol. Evol. 9: 856-871. 107-112. Lee B.-Y. (1998) A phylogenetic study of Apiaceae Heywood V. H. (1982c) Multivariate taxonomic tribe Caucalideae. Ph.D. thesis, University of synthesis of the tribe Caucalideae. In: Cauwet- Illinois at Urbana-Champaign: Urbana. Marc A.-M., Carbonnier J. (eds.) Les Ombel- Lee B.-Y., Downie S. R. (1999) A molecular lif6res. Actes du 2 +me Symposium International phylogeny of Apiaceae tribe Caucalideae and sur les Ombellif6res "Contributions Pluridisci- related taxa: inferences based on ITS sequence plinaires ~ la Systbmatique." Monographs in data. Syst. Bot. 24: 461-479. Systematic Botany from the Missouri Botanical Lid6n M., Fukuhara T., Rylander J., Oxelman B. Garden, vol. 6. Braun-Brumfield, Ann Arbor. (1997) Phylogeny and classification of Fumaria- 727-736. ceae, with emphasis on Dieentra s.l., based on Heywood V. H., Dakshini K. M. M. (1971) Fruit the plastid gene rpsl6 intron. Plant Syst. Evol. structure in the Umbelliferae-Caucalideae. In: 206:411-420. Heywood V. H. (ed.) The Biology and Chemistry Maddison W. P., Maddison D. R. (1992) Mac- of the Umbelliferae, Supplement 1 to the Bo- Clade version 3.0: analysis of phylogeny and tanical Journal of the Linnean Society, Vol. 64. character evolution. Sinauer, Sunderland. Academic Press, New York, 215-232. McNeill J., Parker P. F., Heywood V. H. (1969) A Heywood V. H., Saenz de Rivas C. (1974) Estudio taximetric approach to the classification of the preliminar sobre los Daucus de la Espafia pen- spiny-fruited members (tribe Caucalideae) of the insular. Anal. Inst. Bot. Cavanilles 31: 97-118. flowering-plant family Umbelliferae. In: Cole Holmgren P. K., Holmgren N. H., Barnett L. C. A. J. (ed.) Numerical taxonomy. Academic (1990) Index herbariorum. New York Botanical Press, London, pp. 129-147. Garden, New York. Michel F., Umesono K., Ozeki H. (1989) Compar- Huelsenbeck J. P., Bull J. J., Cunningham C. W. ative and functional anatomy of group II cata- (1996) Combining data in phylogenetic analysis. lytic introns - a review. Gene 82: 5-30. Trends Ecol. Evol. 11: 152-157. Mishler B. D., Albert V. A., Chase M. W., Karis P. Jury S. L. (1986) Fruit and leaf variation in the O., Bremer K. (1996) Character state weighting African species of the Umbelliferae tribe Cau- for DNA restriction site data: asymmetry, calideae. Acta Univ. Ups. Symb. Bot. Ups. 26: ancestors, and the Asteraceae. Cladistics 12: 181-188. 11-19. Katz-Downie D. S., Valiejo-Roman C. M., Teren- Moore D. M. (1971) Chromosome studies in tieva E. I., Troitsky A. V., Pimenov M. G., Lee Umbelliferae. In: Heywood V. H. (ed.) The B.-Y., Downie S. R. (1999) Towards a molecular Biology and Chemistry of the Umbelliferae, B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae 59

Supplement 1 to the Botanical Journal of the Shinozaki K., Ohme M., Tanaka M., Wakasugi T., Linnean Society, Vol. 64. Academic Press, New Hayashida N., Matsubayashi T., Zaita N., York, 233-255. Chunwongse J., Obokata J., Yamaguchi-Shino- Neuhaus H., Scholz A., Link G. (1989) Structure zaki K., Ohto C., Torazawa K., Meng B. Y., and expression of a split chloroplast gene from Sugita M., Deno H., Kamogashira T., Yamada mustard (Sinapsis alba): ribosomal protein gene K., Kusuda J., Takaiwa F., Kato A., Tohdoh rpsl6 reveals unusual transcriptional features N., Shimada H., Sugiura M. (1986) The com- and complex RNA maturation. Curr. Genet. 15: plete nucleotide sequence of the tobacco chlo- 63-70. roplast genome: its gene organization and Olmstead R. G., Palmer J. D. (1992) A chloroplast expression. EMBO J. 5: 2043-2049. DNA phylogeny of the Solanaceae: subfamilial Shneyer V. S., Borschtschenko G. P., Pimenov M. relationships and character evolution. Ann. G. (1995) Immunochemical appraisal of rela- Missouri Bot. Gard. 79: 346-360. tionships within tribe Peucedaneae (Apiaceae). Olmstead R. G., Palmer J. D. (1994) Chloroplast Plant Syst. Evol. 198: 1-16. DNA systematics: a review of methods and data Small E. (1978) A numerical taxonomic analysis of analysis. Amer. J. Bot. 8l: 1205-1224. the Daucus carota complex. Can. J. Bot. 56: 248- Olsen G. J., Matsuda H., Hagstrom R., Overbeek 276. R. (1994) fastDNAml: a tool for construction of St. Pierre M. D., Bayer R. J., Weis I. M. (1990) An phylogenetic trees of DNA sequences using isozyme-based assessment of the genetic vari- maximum likelihood. CABIOS 10: 41-48. ability within the Daucus carota complex (Api- Oxelman B., Lid6n M., Berglund D. (1997) Chlo- aceae: Caucalideae). Can. J. Bot. 68: 2449-2457. roplast rpsl6 intron phylogeny of the tribe Sugiura M. (1992) The chloroplast genome. P1. Sileneae (Caryophyllaceae). Plant Syst. Evol. Mol. Biol. 19: 149-168. 206: 393-410. Swofford D. L. (1993) PAUP: Phylogenetic Anal- Palmer J. D. (1991) Plastid chromosomes: structure ysis Using Parsimony, version 3.1. Computer and evolution. In: Bogorad L., Vasil I. K. (eds.) program distributed by the Illinois Natural The Molecular Biology of Plastids, Vol. 7 in History Survey, Champaign. Vasil, I. K. (editor-in-chief), Cell Culture and Tamamschjan S. G. (1947) Carpological character- Somatic Cell Genetics in Plants. Academic Press, ization of the genus Astrodaucus Drude and New York, pp. 5-53. some Caucasian Caucalinae and Daucinae. So- Pimenov M. G. (1982) Les Ombellif6res d'Asie viet Botany 4: 198-212. Moyenne. In: Cauwet-Marc A.-M., Carbonnier Valiejo-Roman K. M., Pimenov M. G., Terentieva J. (eds.) Les Ombellif~res. Actes du 2&ne Sym- E. I., Downie S. R., Katz-Downie D. S., posium International sur les Ombellif6res "Con- Troitsky A. V. (1998) Molecular systematics of tributions Pluridisciplinaires fi la Systdmatique." the Umbelliferae: using nuclear ribosomal DNA Monographs in Systematic Botany from the internal transcribed spacer sequences to resolve Missouri Botanical Garden, vol. 6. Braun-Brum- issues of evolutionary relationships. Bot. Zhur- field, Ann Arbor. 33-45. nal 83: 1-22. Pimenov M. G., Leonov M. V. (1993) The genera of Wendel J. F., Doyle J. J. (1998) Phylogenetic the Umbelliferae. Royal Botanic Gardens, Kew. incongruence: window into genome history and Plunkett G. M., Downie S. R. (1999) Major lineages molecular evolution. In: Soltis P. S., Soltis D. E., within Apiaceae subfamily Apioideae: a compar- Doyle J. J. (eds.) Molecular systematics of ison of chloroplast restriction site and DNA plants, 2nd edn., Chapman and Hall, New York, sequence data. Amer. J. Bot. 86: 1014-1026. pp. 265-296. Plunkett G. M., Soltis D. E., Soltis P. S. (1996) Williams C. A., Harborne J. B. (1972) Essential oils Evolutionary patterns in Apiaceae: inferences in the spiny-fruited Umbelliferae. Phytochem. based on matK sequence data. Syst. Bot. 21: 11: 1981-1987. 477-495. Wolfe K. H., Morden C. W., Ems S. C., Palmer J. Rieseberg L. H., Soltis D. E. (1991) Phylogenetic D. (1992) Rapid evolution of the plastid trans- consequences of cytoplasmic gene flow in plants. lational apparatus in a nonphotosynthetic plant: Evol. Trends Plants 5: 65-84. loss or accelerated sequence evolution of tRNA 60 B.-Y. Lee and S. R. Downie: Chloroplast DNA Phylogeny of Tribe Caucalideae

and ribosomal protein genes. J. Mol. Evol. 35: Addresses of the authors: Byoung-Yoon Lee and 304-317. Stephen R. Downie (correspondence, E-mail: Zohary M. (1972) Umbelliferae. In: Zohary M. [email protected]), Department of Plant Bi- (ed.) Flora Palaestina, Part 2. Goldberg's Press, ology, University of Illinois at Urbana-Champaign, Jerusalem, pp. 378-460. Urbana, Illinois, 61801 USA.