1358 The evolution of fruit in Scandiceae subtribe Scandicinae (Apiaceae) Krzysztof Spalik, Aneta Wojewódzka, and Stephen R. Downie Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Abstract: Evolutionary relationships among 66 representatives of the family Apiaceae, including 37 species of tribe Scandiceae subtribe Scandicinae, were inferred from separate and combined analyses of fruit morphology and anatomy and nuclear ribosomal DNA internal transcribed spacer (ITS) sequences. Phylogenetic trees inferred from analysis of 35 fruit characters were not congruent to those derived from molecular data and, overall, had much lower bootstrap support values than the latter. Contrary to molecular data, fruit characters did not support the monophyly of subtribe Scandicinae. Fruit data do, however, corroborate the monophyly of nearly every genus within Scandicinae, the affinity of members of the “crown” clade—Anthriscus, Kozlovia (including Krasnovia and Neoconopodium), Geocaryum, Myrrhis, and Osmorhiza—that had been identified in previous molecular analyses, and the sister group relationship between the “crown” clade and the genus Scandix. Phylogenies derived from the analysis of combined ITS and fruit characters were congruent to those inferred from molecular data alone. Reconstructions of ancestral character states using the results of the combined analysis suggest that among extant Scandicinae, the fruits of Athamanta have retained the most plesiomorphic characters. Evolutionary tendencies that have occurred in the fruits of Scandicinae include the broadening of the vascular bundles and vittae, the thickening of the cuticle and epidermal cell wall, the origin of bristles from hairs, the appearance of a pedicel-like appendage, the development of a long beak, and lateral wings. These changes are interpreted as adaptations to fruit dispersal and seed defense. Key words: Apiaceae, Scandiceae subtribe Scandicinae, ITS, fruit morphology. Résumé : Parmi 66 représentants de la famille des Apiaceae, incluant 37 espèces de la tribu des Scandiceae sous-tribu Scandicinae, les auteurs ont déduit les relations évolutives, en utilisant des analyses séparées et combinées de la morphologie et de l’anatomie des fruits, ainsi que des séquences de l’espaceur interne transcrit (ITS) de l’ADN ribosomal nucléique. Les dendrogrammes phylogénétiques, obtenus de l’analyse de 35 caractères des fruits, sont incongrus par rapport à ceux dérivés des données moléculaires et, en général, montrent des valeurs de support en lacet beaucoup plus faibles que ceux-ci. Contrairement aux données moléculaires, les caractéristiques des fruits ne supportent pas la monophylie de la sous-tribu Scandicinae. Cependant, les données sur les fruits corroborent la monophylie d’à peu près tous les genres appartenant aux Scandicinae, l’affinité des membres du clade “crown”—Anthriscus, Kozlovia (incluant Krasnovia et Neoconopodium), Geocaryum, Myrrhis et Osmorhiza—qui ont été identifiés dans des analyses moléculaires précédentes, ainsi que la relation de sororité entre le clade “crown” et le genre Scandix. Les phylogénies dérivées de l’analyse combinée des caractères des ITS et des fruits sont congruentes avec celles obtenues à partir des données moléculaires prises isolément. Les reconstructions des caractères ancestraux, basées sur les résultats d’analyses combinées, suggèrent que parmi les Scandicinae actuelles, les fruits de l’Athamanta ont retenu les caractères les plus plésiomorphes. Les tendances évolutives qui se sont manifestées chez les fruits des Scandicinae incluent l’élargissement des faisceaux vasculaires et des vittae, l’épaississement de la cuticule et de la paroi des cellules épidermiques, l’origine des soies à partir des poils, l’apparence de l’appendice pédicelloïde, le développement d’un long bec, et les ailes latérales. On interprète ces changements comme des adaptations pour la dispersion des fruits et la protection des graines. Mots clés : Apiaceae, Scandiceae sous-tribu Scandicinae, ITS, morphologie des fruits. [Traduit par la Rédaction] Spalik et al. 1374 Introduction Ever since Morison’s (1672) Plantarum umbelliferarum, fruit morphology and anatomy have been regarded as essenReceived March 7, 2001. Published on the NRC Research Press Web site at http://canjbot.nrc.ca on November 16, 2001. K. Spalik1 and A. Wojewódzka. Department of Plant Systematics and Geography, Warsaw University, Aleje Ujazdowskie 4, 00-478 Warsaw, Poland. S.R. Downie. Department of Plant Biology, University of Illinois at Urbana-Champaign, Urbana, IL 61801, U.S.A. 1 Corresponding author (e-mail: spalik@bot.uw.edu.pl). Can. J. Bot 79: 1358–1374 (2001) tial to the taxonomy of Apiaceae (Umbelliferae; Drude 1898; Constance 1971). Despite their general similarity, umbellifer fruits vary with respect to their external and internal features, and nearly all classification systems of the family are based on these characters. Koch (1824), for instance, divided the family into two principal groups, Multiiugatae and Pauciiugatae, on the basis of the number of ribs. He subsequently segregated these groups into 15 tribes upon consideration of fruit shape and compression and characteristics of the ribs and vittae. de Candolle (1830) stressed the importance of endosperm shape, arranging the umbellifers into three groups: Orthospermae, Campylospermae, and Coelospermae. Bentham (1867) gave priority to the number DOI: 10.1139/cjb-79-11-1358 © 2001 NRC Canada Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Spalik et al. of mericarp ribs and two of his three major groups, Haplozygiae and Diplozygiae, are defined by the absence or presence of secondary ribs. Rompel (1895), in his influential study of umbellifer fruit anatomy, emphasized the distribution of calcium oxalate crystals. Drude (1898) considered this character among others when constructing his classification but gave priority to fruit compression and the number of ribs and vittae. For Koso-Poljansky (1916), essential features included the distribution of calcium oxalate crystals, vittae, aerenchyma, and sclerenchyma in the walls of the fruit. The only modern infrafamilial division of the family that was not built upon fruit morphology and anatomy was that proposed by Cerceau-Larrival (1962), who underlined the importance of cotyledon shape and pollen attributes. Although those classification systems that were based on fruit morphology and anatomy used nearly the same set of characters, they differed widely, since the distribution of these characters is largely incongruent. When optimized onto phylogenies obtained using molecular evidence, many of these features are highly homoplastic (Plunkett et al. 1996; Katz-Downie et al. 1999). Contrary to those relationships implied in traditional classification systems, albeit inferred using nonphylogenetic approaches, the phylogenies estimated from diverse sets of molecular data are generally congruent with each other. Several well-supported major lineages have been identified, 10 of which are formally recognized as tribes (Downie et al. 2000b, 2001), with the largest tribe, Scandiceae Spreng., further divided into three subtribes: Scandicinae Tausch, Daucinae Dumort., and Torilidinae Dumort. (Downie et al. 2000a). Scandiceae includes members of former tribes Scandiceae, Caucalideae, and Laserpitieae (sensu Heywood 1971) and with over 120 accessions examined to date is the most extensively sampled major clade of Apiaceae (Lee and Downie 1999, 2000; Downie et al. 2000a). Little regarding fruit evolution in the family is known despite the importance of fruit morphology and anatomy in umbellifer systematics. Some studies, for instance that of Cerceau-Larrival (1962), included hypothetical pathways of fruit evolution. These were, however, based on incorrect assumptions of phylogenetic relationships among taxa (Plunkett et al. 1996; Katz-Downie et al. 1999). The major objective of this paper is to assess the evolutionary history of fruit characters in Scandiceae subtribe Scandicinae. Based on a phylogeny inferred from combined morphological and molecular data sets, we reconstruct fruit morphology and anatomy at internal nodes of this tree and discuss the selective pressures that may have given rise to these changes. Scandicinae constitute a model group for such evolutionary work as molecular and morphological phylogenetic studies indicate that it is a well-defined subtribe (Downie et al. 2000a; Spalik and Downie 2001). Several of its included genera have been recently revised (Engstrand 1977; Lowry and Jones 1984; Spalik 1997). Of the 16 genera recognized in the subtribe as a result of our initial investigations (Downie et al. 2000a), five, Krasnovia, Neoconopodium, Myrrhoides, Tinguarra, and Balansaea, have been since reduced to synonymy (Spalik and Downie 2001; Spalik et al. 2001). The inclusion of Tinguarra in Athamanta, however, needed confirmation from additional data. Spalik and Downie (2001), upon examination of 44 vegetative, inflores- 1359 cence, floral, and fruiting characters in Scandicinae, found those of the fruit to be most congruent with phylogenies inferred from molecular data indicating that phylogenetic signal is retained within these characters. Additional objectives include the examination of the taxonomic utility of fruit anatomy in subtribe Scandicinae to detect those characters that best delimit genera and to determine whether fruit anatomy and morphology confirms the inclusion of Tinguarra in Athamanta. Materials and methods Accessions examined Sixty-six representatives of Apiaceae were included in the cladistic analysis of nuclear ribosomal DNA ITS sequences and fruit morphology and anatomy (Table 1). Scandiceae subtribe Scandicinae was represented by 37 species reflecting all 12 genera recognized in the subtribe on the basis of molecular studies (Downie et al. 2000a, 2000b; Spalik and Downie 2001). Subtribes Daucinae and Torilidinae were each represented by five and three species, respectively. We also considered 21 representatives of tribes Careae, Smyrnieae, and Oenantheae and the apioid superclade (including members of the Angelica, Apium, Heracleum, and Pimpinella subclades). Based on chloroplast DNA evidence, tribe Oenantheae (i.e., Cicuta, Sium, and Oenanthe in this study) is sister to the clade formed from all other aforementioned lineages; however, the relationships among the latter are unresolved (Downie et al. 2000b). We did not consider more distant outgroups (i.e., Heteromorpheae, Bupleureae, Pleurospermeae, and the Komarovia clade), since their ITS sequences are highly divergent; consequently, more positions would have had to be excluded from the analysis because of alignment ambiguities. Moreover, adding such distant and divergent outgroups would actually decrease the consistency of the phylogenetic estimation (Kim 1996). The set of taxa examined in this study is somewhat different from those used in our previous analyses of the subtribe (Downie et al. 2000a; Spalik and Downie 2001), reflecting the availability of specimens with mature fruits. We also added three taxa that were not considered in our prior studies: Conopodium majus, Laserpitium prutenicum, and Todaroa aurea. Voucher information for these three species is provided in Table 1; for all other taxa, references are provided therein where this information has been cited previously. Molecular and morphological data In the molecular analysis, we considered a subset of ITS sequences from our earlier studies (Downie and Katz-Downie 1996, Downie et al. 1998, 2000a; Table 1) plus the three new accessions. Details of the DNA isolation, polymerase chain reaction amplification, and sequencing procedures utilized are provided by Downie et al. (1998, 2000a). The sequences were aligned using CLUSTAL V (Higgins et al. 1992) and manually adjusted. Herbarium material for the morphological analysis was obtained from Jean-Pierre Reduron (Mulhouse, France) and from the following institutions: B, BC, BM, E, ILL, KRA, KRAM, L, MO, P, W, WA (abbreviations according to Holmgren et al. 1990). Mature fruits were selected and their external characters examined using a Nikon SMZ-U dissecting microscope (8–160× magnification). For the anatomical studies, fruits were soaked in water and hand cut or embedded in paraffin and sectioned using a microtome; these dissections were then stained with phloroglucinol or safranin – fast green (Gerlach 1972) and examined using a Nikon Optiphot 2 optical microscope (40–1000× magnification). Drawings were made using a drawing tube or from photographs taken using a Nikon Microflex HFX-DX photographic system. © 2001 NRC Canada 1360 Can. J. Bot Vol. 79, 2001 Table 1. Accessions of Apiaceae examined in this study, GenBank accession numbers for separate ITS 1 and ITS 2 sequences, and the matrix of morphological data. Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Taxon Aegopodium alpestre Ledeb. Aethusa cynapium L. Ammi majus L. Anethum graveolens L. Anthriscus cerefolium (L.) Hoffm. Anthriscus kotschyi Boiss. & Balansa Anthriscus nitida (Wahlenb.) Hazsl. Anthriscus sylvestris (L.) Hoffm. Athamanta cretensis L. Athamanta sicula L. (; Tinguarra sicula (L.) Benth. & Hook. f.) Athamanta turbith (L.) Broth. Bubon macedonicum L. (; Athamanta macedonica (L.) Spreng.) Carum carvi L. Caucalis platycarpos L. Chaerophyllum aromaticum L. Chaerophyllum astrantiae Boiss. & Balansa Chaerophyllum aureum L. Chaerophyllum bulbosum L. Chaerophyllum crinitum Boiss. Chaerophyllum hirsutum L. Chaerophyllum macrospermum (Willd. ex Spreng.) Fisch. & C.A. Mey. Chaerophyllum nodosum (L.) Crantz (; Myrrhoides nodosa (L.) Cannon) Chaerophyllum procumbens (L.) Crantz Chaerophyllum tainturieri Hook. Chaerophyllum temulum L. Cicuta virosa L. Conium maculatum L. Conopodium glaberrimum (Desf.) Engstrand (; Balansaea glaberrima (Desf.) Maire) Conopodium majus (Gouan) Loret Daucus carota L. Falcaria vulgaris Bernh. Foeniculum vulgare Mill. Geocaryum macrocarpum (Boiss. & Spruner) Engstrand Grammosciadium platycarpum Boiss. & Hausskn. Heracleum sphondylium L. Kozlovia capnoides (Decne.) Spalik & S.R. Downie (; Neoconopodium capnoides (Decne.) Pimenov & Kljuykov Kozlovia longiloba (Kar. & Kir.) Spalik & S.R. Downie (; Krasnovia longiloba (Kar. & Kir.) Popov ex Schischk.) Kozlovia paleacea (Regel & Schm.) Lipsky Source of sequence or voucher* GenBank accession No. Morphological characters ITS1 ITS2 1–10 11–20 21–30 31–35 b a b a a d d b d d U78376 U30582 U78386 U30550 U30532 AF073579 AF073595 U79603 AF073685 AF073683 U78436 U30583 U78446 U30551 U30533 AF073580 AF073596 U79604 AF073686 AF073684 0001000000 0011000000 0011000000 0011000000 00120001B1 0012000001 1012000001 1012000001 0011100010 0011100010 0011110000 0011000400 0010000?00 0010000300 1101000001 1100000001 1100000001 1101000001 0013110000 0011110000 5401041132 5201020343 54??030134 2001020001 1320040000 4421040000 3420041000 3420041000 2301021012 340? 011013 11002 10301 ?1101 10310 14031 03011 13032 13022 23131 13221 d d AF073687 AF073541 AF073688 AF073542 0011100010 0011100010 0011110000 0010000000 3301022002 3201121122 23121 11210 b b d d U78377 U78364 AF073631 AF073653 U78437 U78424 AF073632 AF073654 0000000000 1011112120 0010000000 0010000000 0011000000 0013001300 0010000000 0011000000 5401021114 5401020211 2401012213 2201021223 11002 14132 24341 23331 d d d d d AF073655 AF073659 AF073661 AF073665 AF073651 AF073656 AF073660 AF073662 AF073666 AF073652 0010000000 0010000000 0010000000 0010000000 0000000000 0011000000 0011000000 0011000000 0010000000 0012000000 3301021213 3401020213 1400001224 1401012233 ??01012233 24232 13211 04241 14231 13340 d AF073675 AF073676 0000000120 0013000000 2401000333 04241 d d d b b d AF073643 AF073645 AF073641 U78372 U79609 AF073689 AF073644 AF073646 AF073642 U78432 U79610 AF073690 00100000A0 00100000A0 00100000A0 0000000000 0000000000 0011000000 0010000000 0010000000 0011000000 0013000400 0000??0000 0011000000 2401030213 2401020343 2401000213 5401001341 5311041123 3311021012 24121 23222 03221 10202 14002 03111 e a b b d AF336370 U27589 U78378 U78385 AF073607 AF336371 U30315 U78438 U78445 AF073608 0011000000 1011112120 0000000000 0001000000 1012000001 0010AA0000 0011110100 0011000000 0010010000 1011000000 4311021013 520?031003 3401021213 2101021??? 3421041011 14001 ?1101 21011 ?1320 13012 d AF073551 AF073552 0000000000 0011000210 1110022223 11340 a d U30544 AF073601 U30545 AF073602 0001000000 0012A000B1 0010000110 1101000000 2001031000 3320141134 10320 03022 d AF073599 AF073600 10120001B1 1101000000 4320141002 04111 d AF073597 AF073598 1012100121 1101000000 3220142001 04121 © 2001 NRC Canada Spalik et al. 1361 Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Table 1 (concluded). Taxon Laserpitium petrophilum Boiss. & Heldr. Laserpitium prutenicum L. Myrrhis odorata (L.) Scop. Oenanthe pimpinelloides L. Orlaya grandiflora (L.) Hoffm. Osmorhiza berteroi DC. Osmorhiza claytonii (Michx.) C.B. Clarke Osmorhiza depauperata Phil. Osmorhiza mexicana Griseb. Pastinaca sativa L. Petroselinum crispum (Mill.) A.W. Hill Peucedanum cervaria (L.) Lapeyr. Pseudorlaya pumila (L.) Grande Scandix balansae Reut. ex Boiss. Scandix iberica M. Bieb. Scandix pecten-veneris L. Scandix stellata Banks & Sol. Seseli montanum L. Sium latifolium L. Smyrniopsis aucheri Boiss. Smyrnium olusatrum L. Sphallerocarpus gracilis (Bess. ex Trevir.) Koso-Pol. Tinguarra cervariifolia (DC.) Benth. & Hook. f. Tinguarra montana (Webb ex H. Christ) A. Hansen & G. Kunkel Todaroa aurea (Sol.) Parl. Tommasinia verticillaris (L.) Bertol. Torilis nodosa (L.) Gaertn. Torilis trichosperma (L.) Spreng. (; Chaetosciadium trichospermum (L.) Boiss.) Source of sequence or voucher* d GenBank accession No. Morphological characters ITS1 AF073567 ITS2 AF073568 1–10 0001111020 11–20 0011100?00 21–30 ????022012 31–35 ?21?1 f a b a b d AF336374 U30530 U78371 U30524 U78365 AF073615 AF336375 U30531 U78431 U30525 U78425 AF073616 0001111020 0012100021 0000000000 1011112020 1112100021 1112100021 0011000000 0111110000 0010000400 0011000C00 1101000000 1101000000 3001021003 2421042122 430?002343 3101022104 1401040120 1401041133 13201 13041 ?030? 13230 22141 23141 d d a b c a b d a d a b b a d AF073611 AF073621 U30546 U78387 AF008608 U30522 U79621 AF073627 U30538 AF073629 U30578 U78370 U78393 U30594 AF073677 AF073612 AF073622 U30547 U78447 AF009087 U30523 U79622 AF073628 U30539 AF073630 U30579 U78430 U78453 U30595 AF073678 1112100021 1112100021 0001000000 0001000000 0001000000 0001112020 0020100020 0020100020 0020100120 0021100120 0001100010 0001000000 0001000000 0001000000 0001000000 1101000000 1101000000 0010000110 0011000000 0010000010 0011000100 1111000000 1113000000 1112000000 11A0000000 0010000000 001?010400 0011010000 0011110? 00 0011110000 1511040122 1511040132 1001031000 5301031013 3001021000 3001032114 1401020223 0411021222 0311021132 0401040122 5201031133 5301011243 5421022133 5511032012 4401030022 12041 22041 10320 21102 13320 13230 04242 14341 04241 13241 10100 10002 24122 24032 14012 d AF073681 AF073682 0011100010 0013110000 2401022232 13001 d AF073679 AF073680 0001100010 0011A10000 3200021040 13021 g c b a AF336372 AF008609 U30534 U78363 AF336373 AF009088 U30535 U78423 0001100010 0001000000 0001112120 0001113011 0011000010 0010000010 1011001001 1011001000 3111?1?241 3001?32001 5320040001 3310032003 13031 ??320 1300? 03111 Note: Synonymy is provided for those species that have recently changed generic placement. With the exception of three species, whose voucher information is provided herein, source of ITS data and voucher information and deposition are available in several previous publications. Characters are described in Table 2. Question marks denote missing data, and A, B, and C denote character states “0 or 1”, “0 or 2”, and “1 or 2”, respectively. *Sources of ITS sequences or voucher specimens are as follows: a, Downie and Katz-Downie (1996); b, Downie et al. (1998); c, Katz-Downie et al. (1999); d, Downie et al. (2000a); e, England, South Essex, Stock, 25 May 1965, Walters & Sell (WA); f, Poland, Klaudyn (near Pruszków), 2 August 1971, Nowak (WA); g, cultivated in Conservatoire Botanique Mulhouse (France) no. 95183, ex Tenerife, Barranco del Infierno, 27 September 1996, Reduron (WA). Twenty qualitative (Nos. 1–20, Tables 1 and 2) and 15 quantitative (Nos. 21–35) characters were examined. Of the qualitative characters, 13 refer to the external morphology of the fruit with all but one having been considered in a previous study (Spalik and Downie 2001). The seven remaining qualitative characters, representing fruit anatomy, are new. The quantitative characters are also new for this study, and were coded as binary or multistate-ordered characters arbitrarily after examining their distributions. In total, 15 characters (Nos. 3, 14, 21–23, and 26–35) were treated as ordered. All data matrices and PAUP nexus files (Swofford 1998) used in this study can be obtained from the authors upon request. Phylogenetic inference Morphological and ITS data were analyzed separately and combined using both maximum parsimony and distance methods available in PAUP* version 4.0. To account for the varying number of states for the morphological characters, ranging from two to five, all characters were weighted using PAUP’s scaling option and fractional weights were employed. These weights were used in both the separate analysis of morphological data as well as in the combined analysis; for the latter, each ITS position was assigned a weight of 1. The maximum parsimony analysis included heuristic searches with 500 random addition replicates and tree bisection– reconnection branch swapping. Distance methods included neighbor-joining and others depending upon the data being analyzed: Jukes–Cantor and Kimura two-parameter methods (Kimura 1980) for the ITS data set and mean and total character differences for both the morphological and combined data sets. Bootstrap support (Felsenstein 1985) was estimated using 1000 resampled data sets; in the maximum parsimony analysis, heuristic searches with simple addition sequence of taxa were employed, saving a maximum of 10 trees from each replicate. All trees were rooted with the three included members of tribe Oenantheae. © 2001 NRC Canada 1362 Can. J. Bot Vol. 79, 2001 Table 2. Fruit characters used in the cladistic analysis of anatomical and morphological data. Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Character No. Character States Comments 1 Crown of scales (bristles) at fruit base 0, absent; 1, present 2 Pedicel-like appendage 0, absent; 1, present 3 Beak [0, obsolete : 1, relatively short] : 2, long 4 Primary ridge shape 0, broad rounded or obsolete; 1, filiform or winged; 2, angular (sometimes only at the top of the fruit) 5 Primary ridge indumentum 0, absent; 1, present 6 Secondary ridges 0, absent; 1, present Scales (bristles) forming the crown on the pedicel at the base of fruit, if present, are distinctly longer than those that may occur along the pedicel; the presence of this crown is generally characteristic for “crown” Scandicinae, although in some species it may be reduced to single scales or even be absent; outside the subtribe, this character also occurs in Daucinae This appendage is formed when the seed does not fill the lower part of the fruit; it is synapomorphic for Osmorhiza, with the exception of O. occidentalis (not considered in the present study); its absence in the latter is considered to be a reversal (Spalik and Downie 2001) Similar to the pedicel-like appendage, a distinct beak is formed when the seed does not fill the fruit entirely; a short beak occurs in many independent lineages and may also be lost repeatedly (as in some species of Chaerophyllum); a long beak is synapomorphic for Scandix This typology is simplified, accounting for variation seen in Scandicinae rather than the entire family; it is based on the ridge shape seen in transverse section of a dried mericarp; narrow obtuse or winged ridges that are clearly delimited from valleculae were scored as filiform; broadly triangular ridges that occur in a pentagonal or star-shaped pattern in “crown” Scandicinae were scored as angular; these ridges are not clearly delimited from valleculae since they are formed by the edges of the pentangle; sometimes they are extant only in the upper part of the fruit (as in Anthriscus and Kozlovia); however, after soaking fruits in water, filiform and angular ridges are not that easy to distinguish; therefore, the filiform appearance of ridges may result from the drying and contraction of the pericarp in the valleculae, whereas in the ridges it is reinforced by the vascular bundles; in the “crown” species of Scandicinae, the valleculae are only fairly contracted because of the thickened cuticle and cell walls of the epidermis; similarly, broad rounded ridges, exemplary of Chaerophyllum, possess valleculae (meaning in Latin “small valleys”) that are contracted when dried; yet after soaking, valleculae may be higher than the ridges (compare drawings of entire dried mericarps with the sections drawn from soaked fruits, Fig. 4). In Scandicinae, the indumentum may cover the entire fruit or be confined to the primary ridges; primary ridge indumentum refers only to those hairs or bristles that are distinctly lined along the ridges Pronounced secondary ridges occur only in members of Daucinae and Torilidinae 7 Secondary ridge appendages Tubercles at fruit surface 0, absent; 1, wings; 2, spines; 3, hairs 0, absent; 1, present 8 A tubercle usually forms the base of a bristle, although the latter may be reduced to a short tooth or be completely absent (as in Kozlovia longiloba) © 2001 NRC Canada Spalik et al. 1363 Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Table 2 (continued). Character No. 9 Character Fruit indumentum States 0, absent; [1, hairs; 2, bristles] 10 Cuticle appearance 0, dull; 1, shiny 11 Cuticle texture 0, smooth or striate; 1, aculeate 12 Epidermis 0, uniform in color (not areolate); 1, areolate (scalariform) 13 Vallecular vittae at fruit maturity No. of intrajugal vittae per ridge No. of vallecular vittae per vallecula 0, obsolete or compressed; 1, extant (not compressed) 0, none : 1, one : 2, two : 3, greater than two 0, one; 1, more than one (sometimes anastomosing) No. of commissural vittae Position of commissural vittae 0, two; 1, more than two (sometimes anastomosing) 0, between commissural bundles; 1, below bundles (between the bundle and the endosperm) 0, absent; 1, ring around endosperm; 2, above vallecular vittae; 3, above vascular bundles and vallecular vittae; 4, between vascular bundles 0, not winged; 1, winged 14 15 16 17 18 Fruit sclerification (apart from vascular bundles) 19 Lateral primary ridges 20 Position of commissural bundles Ratio of mericarp width (parallel to commissure) to its length 21 22 Ratio of mericarp width (parallel to commissure) to its thickness 23 Cuticle and cell-wall thickness at cell center 0, distinctly lateral; 1, close to the carpophore 0, smaller than 0.05 : 1, 0.05– 0.10 : 2, 0.11–0.14 : 3, 0.15– 0.20 : 4, 0.21–0.26 : 5, greater than 0.26 0, smaller than 0.3 : 1, 0.31–0.45 : 2, 0.46–0.60 : 3, 0.61–0.79 : 4, 0.80–1.00 : 5 greater than 1.00 [0, smaller than 9 µm : 1, 9–13 µm] : 2, greater than 13 µm Comments Several species are polymorphic and include members with glabrous or hairy (bristled) fruits; in Kozlovia longiloba, tubercles usually lack bristles but sometimes end with short hyaline teeth that appear homologous to bristles This character is difficult to determine precisely in badly preserved herbarium specimens or when the fruits are immature; a shiny fruit appearance could be related to the thickness of the cuticle and cell wall (character No. 23); however, these characters do not exactly coincide; for example, members of Osmorhiza have shiny fruits, while their cuticles are not very thick An aculeate cuticle texture results from the formation of small projections above the center of the epidermal cells; in Scandix, they are only found close to the fruit base, with the exception of Scandix stellata, where the entire fruit is aculeate; the sizes of these projections are considered in character No. 24 An areolate epidermal appearance is due to thickened transverse cell walls (Spalik 1997); it is not always well developed and sometimes occurs only on part of the fruit Intrajugal vittae are small secretory canals that are situated in the primary ribs above the vascular bundles This character, as well as the subsequent one, is difficult to assess in those species in which the vittae are compressed at fruit maturity Commissural vittae situated below the commissural bundles were found only in those examined species of Torilidinae Apart from the vascular bundles, different parts of the fruit may be lignified; sclerenchymatic cells may form a closed ring in the endocarp around the endosperm or be present in different parts of the mesocarp Among Scandicinae, only Todaroa aurea has winged lateral primary ribs In Scandicinae, there is no boundary between the cell wall and the cuticle; in those species with thick cuticles, the transverse cell walls of the epidermal cells are also thickened, although not as much as the outer walls; the thickening of transverse cell walls results in an areolate appearance of the epidermis (character No. 12), although the latter is also found in Scandix and Osmorhiza which have cuticles not exceeding 13 µm in thickness © 2001 NRC Canada 1364 Can. J. Bot Vol. 79, 2001 Table 2 (concluded). Character No. 24 Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. 25 26 27 Character Ratio of cuticle and cell-wall thickness at cell side to that at cell center Angle of styles States 0, smaller than 0.9; 1, equal to or greater than 0.9 Comments 0, smaller than 280°; 1, equal to or greater than 280° Percent girth not protected by thickened bundles and vittae (in transverse section of mericarp) Style length [0, smaller than 10% : 1, 11–20% : 2, 21–50% : 3, 51–69%] : 4, equal to or greater than 70% The angle of the style is a highly variable character even within the same species; moreover, it is also influenced by fruit maturity and drying conditions; some species, particularly members of Kozlovia, have styles that are bent at their base and directed downwards so that they touch the fruit A transverse section through the mericarp shows vascular bundles alternating with vittae to form a ring that protects the endosperm; this character accounts for the proportion of “gaps” in this ring 0, smaller than 1 mm : 1, 1–2 mm : 2, greater than 2 mm [0, smaller than 0.0025 : 1, 0.0025–0.006] : [2, 0.0061– 0.025 : 3, greater than 0.025] 28 Ratio of dorsal bundle size to mericarp size in transverse section 29 Ratio of dorsal bundle thickness to its width (in transverse section of mericarp) Ratio of lateral bundle size to commissural bundle size (in transverse section of mericarp) Ratio of mean epidermis cell width in valleculae to that in ridges Ratio of endosperm furrow depth to endosperm thickness 0, smaller than 0.027 : 1, 0.0271–0.047 : 2, 0.0471– 0.062 : 3, 0.0621–0.1 : 4, greater than 0.1 0, smaller than 0.45 : [1, 0.451– 0.65 : 2, 0.651–0.78 : 3, 0.781–0.95 : 4, greater than 0.95] 33 Ratio of commissure width to mericarp width 0, smaller than 0.38 : 1, 0.39– 0.6 : 2, 0.61–0.9 : 3, greater than 0.9 34 Mean fruit length 35 Ratio of endosperm thickness to its width 0, smaller than 4 mm : 1, 4–4.9 mm : 2, 5–6.9 mm : 3, 7–8.9 mm : 4, equal to or greater than 9 mm 0, smaller than 0.5 : 1, 0.5–0.8 : 2, greater than 0.8 30 31 32 This variable quantifies the relative bundle size; within Scandicinae, enlarged bundles are characteristic of Chaerophyllum and several members of Scandix; generally, however, this character is homoplastic; to simplify calculations, the bundles and mericarp transverse sections described in character Nos. 28–30 are treated as rectangles and their sizes calculated as width by thickness Enlarged commissural bundles are characteristic of Anthriscus although they are not unique to this genus 0, up to 1 : 1, 1.01–2 : 2, greater than 2 [0, smaller than 0.01 : 1, 0.01– 0.11] : [2, 0.111–0.14 : 3, 0.141–0.27 : 4, greater than 0.27] Although also present in other lineages of Apiaceae, the endosperm furrow at the commissural face well defines the branch of Apiaceae comprising Scandiceae; in some members of Scandiceae with dorsally flattened mericarps (e.g., Daucus carota), the value of this character may be quite low Usually denoted as “commissure broad versus constricted,” this character is often used in umbellifer taxonomy, although it may be quite variable even within the same genus Flattening of the endosperm does not necessarily correlate with the flattening of the fruit; the latter may arise by the broadening of lateral ridges; hence, the shape of the endosperm remains unchanged Note: Double-headed arrows indicate ordered characters. States given in brackets were subsequently merged to reduce homoplasy (see text for further discussion). © 2001 NRC Canada Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Spalik et al. The pattern of evolution of each fruit character was assessed using MacClade version 3.7 (Maddison and Maddison 1992) and the minimal length trees resulting from maximum parsimony analysis of combined morphological and molecular data sets. For each character, retention (RI), consistency (CI), and rescaled consistency (RC) indices were calculated. For each of these indices, nearly all multistate-ordered characters had very low values; thus, some states were subsequently merged to reduce homoplasy (see Table 2). By optimizing the morphological characters onto the strict consensus tree resulting from the combined analysis, we identified those characters that are most useful in delimiting well supported clades in Scandicinae. Based on the reconstruction of character states at the internal nodes of the tree, as estimated by MacClade, we attempted to hypothesize fruit morphology and anatomy of the ancestors of all genera of Scandicinae, of some specific groups of genera, and of the entire subtribe. In those cases when MacClade provided equivocal reconstructions, we usually assumed that the loss of a character is more probable than its independent gain. However, we sometimes chose other reconstructions of ancestral states than those suggested by MacClade. For instance, Spalik (1996) noted that in Anthriscus bristles were lost several times so we therefore assumed that the loss of indumentum is more probable than its gain. Results Phylogenetic analyses Equally weighted maximum parsimony analysis of all included ITS positions resulted in 30 minimal length trees, each of 1539 steps, with CIs of 0.425 and 0.394 (with and without uninformative characters, respectively) and a RI of 0.725. The strict consensus of these trees was well resolved (Fig. 1) with only few polytomies occurring in terminal branches. Distance trees (not shown) were generally similar in topology to those derived from the maximum parsimony analysis, and all trees were congruent to those obtained using a broader sampling of Scandicinae and outgroup genera (Downie et al. 2000a). In all analyses herein, tribe Scandiceae is strongly supported as monophyletic and comprises three major lineages corresponding to subtribes Scandicinae, Daucinae, and Torilidinae. Within Scandicinae, several distinct branches are identified, with eight of these equivalent to the following currently recognized genera (sensu Spalik and Downie 2001): Anthriscus, Kozlovia (including Krasnovia and Neoconopodium), Geocaryum, Myrrhis, Osmorhiza, Scandix, Chaerophyllum (including Myrrhoides), and Sphallerocarpus. Sister to this clade of eight genera is a polytomous clade, supported by a 97% bootstrap value, comprising Athamanta, Conopodium (including Balansaea), Tinguarra, and Todaroa. However, neither Tinguarra nor Conopodium is maintained as monophyletic. In a subset of the 30 trees obtained from the parsimony searches (not shown), both species of Tinguarra fell together. In contrast, the two examined members of Conopodium never formed a clade; in all trees, Conopodium majus is sister to Athamanta, while Conopodium glaberrimum, formerly recognized in the monotypic genus Balansaea, is placed one node away. The genera Anthriscus, Kozlovia, Geocaryum, Myrrhis, and Osmorhiza formed a strongly supported clade (99% bootstrap value) and has been collectively recognized as “crown” Scandicinae (Spalik and Downie 2001). Within this “crown” clade, Anthriscus and Kozlovia are immedi- 1365 ately related, but this affinity is only poorly supported (bootstrap value 28%). The genus Kozlovia is supported only moderately with a bootstrap value of 72%. Apart from its type, Kozlovia paleacea, this genus includes those members formerly treated in Krasnovia and Neoconopodium (Spalik and Downie 2001). Although the bootstrap value for Anthriscus is low (44%), the monophyly of this genus was confirmed previously by separate analyses of morphological data (Spalik and Downie 2001). In all molecular analyses, Scandix is sister to “crown” Scandicinae. Phylogenetic analyses of the 35 fruit characters resulted in trees that were not congruent to those derived from ITS sequences. In the neighbor-joining analysis of total character differences, neither Scandiceae nor any of its three subtribes is maintained as monophyletic (Fig. 2). Members of subtribes Torilidinae and Daucinae fell together, with the former polyphyletic and arising from the latter. Members of subtribe Scandicinae occur in six different clades. In contrast, the monophyly of most genera was confirmed, but only Scandix and Osmorhiza received bootstrap support values higher than 50%. Members of “crown” Scandicinae formed a clade sister to Scandix, and within the former Anthriscus and Kozlovia ally as sister taxa. The infrageneric relationships within Scandix, Osmorhiza, Kozlovia, and Anthriscus were also identical or consistent to those inferred from ITS data. The close relationship among Athamanta, Tinguarra, and Todaroa, as inferred by the molecular analyses, was confirmed in the neighborjoining tree. However, in those trees inferred using maximum parsimony (not shown), this clade did not occur. Instead, Todaroa aurea and Tinguarra montana allied with Seseli montanum and Bubon macedonicum. Within Athamanta, Athamanta sicula is sister to the two other species, in accordance with the molecular analyses. Conopodium is supported as monophyletic only in the neighbor-joining tree, but it is placed distantly from other members of Scandicinae. In the maximum-parsimony trees (not shown), this genus forms a grade at the base of the “crown” clade. Chaerophyllum macrospermum groups with Grammosciadium platycarpum in all trees inferred from morphological data, with this clade either associated with members of the apioid superclade in the neighbor-joining tree (Fig. 2) or with other Chaerophyllum species in the maximum parsimony trees (not shown). Chaerophyllum nodosum, previously recognized in the monotypic genus Myrrhoides (= Physocaulis) and recently reinstated in Chaerophyllum (Spalik and Downie 2001), consistently fell within the latter. In all trees, Chaerophyllum (save Chaerophyllum macrospermum) is sister to Falcaria. Maximum parsimony analysis of combined ITS and differentially weighted fruit characters resulted in two minimal length trees, each of length 1766.4 steps, CIs of 0.390 and 0.362 (with and without uninformative characters, respectively), and a RI of 0.704. The topology of their strict consensus (Fig. 3) was nearly identical to that derived from parsimony analysis of ITS data (Fig. 1) but with increased resolution among the basal branches of Scandicinae and generally higher bootstrap values overall. Similar results were obtained using the neighbor-joining method. In all analyses of combined data, Conopodium and Tinguarra are each monophyletic with the latter sister to Athamanta. © 2001 NRC Canada 1366 Can. J. Bot Vol. 79, 2001 Fig. 1. Strict consensus of 30 minimal-length 1539-step trees derived from equally weighted maximum parsimony analysis of 66 nuclear ribosomal DNA ITS sequences from Scandiceae subtribe Scandicinae and outgroups (CI = 0.394 excluding uninformative characters; RI = 0.725). Bootstrap values for 1000 replicate analyses are indicated in percent along respective nodes; only those compatible with the majority-rule consensus tree are indicated. Brackets indicate genera recognized in subtribe Scandicinae or tribal, subtribal and informal clade divisions in Apiaceae identified by Downie et al. (2000a, 2001). Names of informal groups are identified by double quotes to distinguish them from formally recognized taxa. Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Maximum Parsimony Strict Consensus Tree 28 66 72 41 47 99 64 94 100 96 100 100 99 87 74 64 100 76 88 89 99 87 67 90 97 81 63 97 100 83 84 92 95 95 98 84 75 92 100 61 91 100 45 77 82 100 100 95 100 100 96 69 59 100 Anthriscus nitida Anthriscus sylvestris Anthriscus kotschyi Anthriscus cerefolium Kozlovia paleacea Kozlovia longiloba Kozlovia capnoides Geocaryum macrocarpum Myrrhis odorata Osmorhiza depauperata Osmorhiza mexicana Osmorhiza berteroi Osmorhiza claytonii Scandix iberica Scandix pecten-veneris Scandix balansae Scandix stellata Chaerophyllum astrantiae Chaerophyllum macrospermum Chaerophyllum crinitum Chaerophyllum aromaticum Chaerophyllum bulbosum Chaerophyllum aureum Chaerophyllum procumbens Chaerophyllum tainturieri Chaerophyllum temulum Chaerophyllum hirsutum Chaerophyllum nodosum Sphallerocarpus gracilis Athamanta cretensis Athamanta turbith Athamanta sicula Conopodium majus Conopodium glaberrimum Tinguarra cervariifolia Tinguarra montana Todaroa aurea Daucus carota Pseudorlaya pumila Laserpitium prutenicum Orlaya grandiflora Laserpitium petrophilum Torilis nodosa Torilis trichosperma Caucalis platycarpos Smyrnium olusatrum Seseli montanum Tommasinia verticillaris Peucedanum cervaria Aethusa cynapium Smyrniopsis aucheri Heracleum sphondylium Pastinaca sativa Conium maculatum Bubon macedonicum Ammi majus Petroselinum crispum Anethum graveolens Foeniculum vulgare Grammosciadium platycarpum Carum carvi Falcaria vulgaris Aegopodium alpestre Cicuta virosa Sium latifolium Oenanthe pimpinelloides Anthriscus Kozlovia Geocaryum Myrrhis Osmorhiza Scandix Chaerophyllum Scandiceae 44 Scandicinae 100 ITS sequence Sphallerocarpus Athamanta Conopodium Tinguarra Todaroa Daucinae Torilidinae Smyrnieae "Angelica" "Heracleum" "Pimpinella" "Apium" Careae Oenantheae © 2001 NRC Canada Spalik et al. 1367 Fig. 2. Neighbor-joining tree inferred from 35 morphological and anatomical fruit characters for 66 representatives of Scandiceae subtribe Scandicinae and outgroups. Branch lengths are proportional to distances estimated using total character differences (note scale bar). Percent bootstrap values for 1000 replicate analyses are indicated along the nodes for those groups that occurred in the majorityrule consensus tree. Members of Scandicinae are bracketed in the far right of the figure. Anthriscus nitida Anthriscus sylvestris 44 Anthriscus cerefolium Total Character Difference Anthriscus kotschyi 59 Neighbor-Joining Tree 95 Kozlovia longiloba 40 Kozlovia paleacea 0 1 2 3 4 5 35 Kozlovia capnoides Distance Osmorhiza depauperata 81 65 Osmorhiza mexicana 28 99 Osmorhiza berteroi 25 Osmorhiza claytonii Geocaryum macrocarpum Myrrhis odorata 19 55 Scandix iberica 37 Scandix pecten-veneris 68 Scandix balansae Scandix stellata 25 Caucalis platycarpos Daucus carota 12 54 Torilis trichosperma Torilis nodosa 15 38 Laserpitium petrophilum 80 Orlaya grandiflora 45 Pseudorlaya pumila Laserpitium prutenicum 67 Athamanta cretensis 32 Athamanta turbith 26 Athamanta sicula 24 Tinguarra cervariifolia 8 11 Tinguarra montana Todaroa aurea 62 Bubon macedonicum Seseli montanum 45 Chaerophyllum aromaticum Chaerophyllum hirsutum 3 Chaerophyllum astrantiae 35 Chaerophyllum aureum Chaerophyllum procumbens 35 Chaerophyllum tainturieri 6 Chaerophyllum crinitum 22 Chaerophyllum nodosum Chaerophyllum temulum Chaerophyllum bulbosum Falcaria vulgaris 99 Heracleum sphondylium 6 50 Pastinaca sativa 78 Peucedanum cervaria 60 Tommasinia verticillaris 51 Anethum graveolens 11 Foeniculum vulgare Chaerophyllum macrospermum 34 Grammosciadium platycarpum Smyrnium olusatrum 26 Sphallerocarpus gracilis Aegopodium alpestre 14 Conium maculatum Smyrniopsis aucheri 31 Carum carvi 26 Petroselinum crispum 40 Conopodium glaberrimum Conopodium majus Ammi majus Aethusa cynapium 48 Cicuta virosa Oenanthe pimpinelloides Sium latifolium 98 Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Fruit characters 36 Anthriscus Kozlovia Osmorhiza Geocaryum Myrrhis Scandix Daucinae + Torilidinae Athamanta Tinguarra Todaroa Chaerophyllum Chaerophyllum Sphallerocarpus Conopodium © 2001 NRC Canada 1368 Can. J. Bot Vol. 79, 2001 Fig. 3. Strict consensus of two minimal length trees, each of length 1766.4 steps, inferred from maximum parsimony analysis of combined morphological and ITS sequence data for 66 representatives of Scandiceae subtribe Scandicinae and outgroups (CI = 0.362 excluding uninformative characters; RI = 0.704). Numbers along nodes denote bootstrap values for 1000 replicate analyses; only those compatible with the majority-rule consensus tree are indicated. Those morphological characters and states most useful in delimiting genera and suprageneric lineages are indicated; character numbers refer to those of Tables 1 and 2. Brackets are the same as described in Fig. 1. Fruit Characters 2. Pedicel-like appendage: 0, absent; 1, present 3. Beak: 0, obsolete or short; 1, long 4. Primary ridges: 0, broad; 1, filiform; 2, angular 6. Secondary ridges: 0, absent; 1, present 9. Fruit indumentum: 0, absent; 1, present; 2, polymorphic 10. Cuticle appearance: 0, dull; 1, shiny 11. Cuticle texture: 0, smooth or striate; 1, aculeate 12. Epidermis: 0, uniform in color; 1, areolate 13. Vallecular vittae at fruit maturity: 0, obsolete; 1, extant; 2, polymorphic 17. Position of commissural vittae: 0, between bundles; 1, below bundles 20. Position of commissural bundles: 0, lateral; 1, close to the carpophore 23. Cuticle and cell wall thickness: 0, up to 13 µm; 1, more than 13 µm 25. Angle of styles: 0, less than 280°; 1, no less than 280° 26. Percent girth not protected: 0, less than 70%; 1, no less than 70% 28. Ratio dorsal bundle/mericarp: 0, up to 0.006; 1, more than 0.006 30. Ratio lateral/commissural bundle: 0, up to 0.45; 1, more than 0.45 32. Ratio endosperm furrow/thickness: 0, less than 0.11; 1, no less than 0.11 99 73 89 93 41 States: 0, ; 1, ; 2, 9 95 2 3 4 6 9 10 11 12 13 17 20 23 25 26 28 30 32 91 32 86 95 100 20 30 Anthriscus Kozlovia Geocaryum Myrrhis Osmorhiza Scandix Chaerophyllum Scandiceae Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Maximum Parsimony Strict Consensus Tree 9 Anthriscus nitida Anthriscus sylvestris Anthriscus cerefolium 9 9 13 44 Anthriscus kotschyi Kozlovia longiloba 25 94 9 43 89 Kozlovia paleacea 23 Kozlovia capnoides 9 12 58 Geocaryum macrocarpum 11 4 10 Myrrhis odorata 100 94 Osmorhiza depauperata 96 Osmorhiza mexicana 2 13 30 100 Osmorhiza berteroi 11 12 26 Osmorhiza claytonii 28 100 Scandix iberica 4 26 100 Scandix pecten-veneris 28 3 100 Scandix balansae 13 Scandix stellata 65 Chaerophyllum aromaticum 46 Chaerophyllum bulbosum Chaerophyllum aureum 100 90 Chaerophyllum astrantiae 73 Chaerophyllum macrospermum 9 49 Chaerophyllum crinitum 99 Chaerophyllum procumbens 9 86 4 28 Chaerophyllum tainturieri 93 62 Chaerophyllum temulum Chaerophyllum hirsutum Chaerophyllum nodosum Sphallerocarpus gracilis 99 Athamanta cretensis 93 Athamanta turbith 40 28 Athamanta sicula 25 Tinguarra cervariifolia Tinguarra montana 30 9 93 Conopodium glaberrimum Conopodium majus 28 Todaroa aurea 32 100 Daucus carota 83 Pseudorlaya pumila 89 Laserpitium prutenicum 6 93 Orlaya grandiflora Laserpitium petrophilum 10 11 99 6 17 20 23 26 Torilis trichosperma 85 Torilis nodosa 28 Caucalis platycarpos 3032 Smyrnium olusatrum Peucedanum cervaria 66 96 Tommasinia verticillaris 9 100 28 Seseli montanum 45 Aethusa cynapium 23 32 Smyrniopsis aucheri 84 30 Heracleum sphondylium 100 57 43 Pastinaca sativa 4 13 26 32 Conium maculatum 9 25 Bubon macedonicum Ammi majus 100 100 Petroselinum crispum Anethum graveolens 100 Foeniculum vulgare 4 100 28 Carum carvi 4 28 Grammosciadium platycarpum 96 62 Falcaria vulgaris 26 Aegopodium alpestre 4 Cicuta virosa 28 56 100 Oenanthe pimpinelloides Sium latifolium 65 Scandicinae ITS and Fruit Characters Sphallerocarpus Athamanta Tinguarra Conopodium Todaroa Daucinae Torilidinae Smyrnieae "Angelica" "Heracleum" "Pimpinella" "Apium" Careae Oenantheae © 2001 NRC Canada Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Spalik et al. Taxonomic utility of fruit characters The distribution of 17 morphological characters, nine of which had RC ≥ 0.37 (character Nos. 2, 3, 6, 9, 10, 12, 13, 17, and 20), is highlighted on the strict consensus tree resulting from maximum parsimony analysis of combined data (Fig. 3). These characters were emphasized because of their ability in defining monophyletic groups and their lower homoplasy relative to the other characters examined. Ten of these characters describe external fruit morphology. A pedicel-like appendage (character No. 2) is unique to Osmorhiza, while a long beak (character No. 3) is synapomorphic for Scandix. An aculeate and areolate fruit epidermis (character Nos. 11 and 12) unites Scandix and all members of “crown” Scandicinae. The latter are also characterized by angular primary ridges (character No. 4) and shiny cuticles (character No. 10), with some possessing a thickened cuticle (character No. 23) and vittae that are compressed at fruit maturity (character No. 13). Commissural bundles situated close to the carpophore (character No. 20) are characteristic for Anthriscus, but such bundles also occur outside of the subtribe in Torilis nodosa. Styles that are bent downwards (character No. 25) distinguish Kozlovia from closely related Anthriscus but are also found in Bubon macedonicum. A fruit indumentum, whether comprised of hairs or bristles (character No. 9), is generally characteristic for the entire tribe; however, in Conopodium and most species of Chaerophyllum the fruits are glabrous. Secondary ridges (character No. 6) with varying types of appendages (character No. 7; not shown) occur in Daucinae and Torilidinae; those examined members of the latter are also characterized by commissural vittae situated below the vascular bundles (character No. 17). In fruits of both of Daucinae and Torilidinae, as well as of some other lineages of Apiaceae, different forms of sclerification (character No. 18; not shown) may occur. This sclerification, however, is absent in subtribe Scandicinae. Despite somewhat higher homoplasy, four other characters are also useful in defining monophyletic groups. The entire tribe as well as its sister group Smyrnieae is characterized by a grooved endosperm (character No. 32); however, this state also occurs in Smyrniopsis aucheri, Peucedanum cervaria, and Conium maculatum, while some members of Scandiceae reverted to an almost flat endosperm (e.g., Daucus carota). Commissural bundles are apparently much larger than lateral ones (character No. 30) in all members of Anthriscus but this feature also occurs in several other species. The size of the vascular bundles is exemplified by the ratio of dorsal bundle size to the size of the entire transverse section of the mericarp (character No. 28). Scandicinae generally have small vascular bundles, with exceptions including Chaerophyllum and two species of Scandix. Somewhat enlarged dorsal bundles characterize Tinguarra cervariifolia and Todaroa aurea. Outside the subtribe, this character is quite homoplastic. The vascular bundles and intervening vallecular vittae form a more or less continuous shield that surrounds the endosperm. However, in those members of the “crown” clade and in Scandix stellata, vascular bundles and vittae are reduced in size; hence, more than 70% of endosperm girth (character No. 26) is not protected. A similar pattern exists in Conium maculatum, Aegopodium alpestre, and Torilis nodosa. 1369 It is noteworthy that nearly all genera of subtribe Scandicinae may be unambiguously defined based on fruit characters (Figs. 3 and 4). The only exception is Tinguarra, for which we did not find any characters to distinguish it from its sister Athamanta. Discussion Taxonomic utility of fruit characters Umbellifer fruits exhibit an outstanding array of morphological and anatomical modifications, many of which are believed to constitute adaptations for various modes of seed dispersal. These characters are therefore susceptible to convergence and may constitute poor indicators of monophyletic groups. As an example, tribe Peucedaneae was defined on the basis of a distinct dorsal flattening of the mature fruit with its lateral ridges expanded into winglike appendages. Differences in the morphology of these appendages served to divide the tribe into three subtribes: Angelicinae, Ferulinae (Peucedaninae), and Tordyliinae (Drude 1898). However, as Theobald (1971) pointed out, it is easy to picture the evolution of dorsal flattening and wing formation as a dispersal mechanism in many independent lineages derived from less specialized types, and this is indeed what Downie et al. (2000c) inferred from their molecular analyses. Fruit characters have also proved limiting for phylogenetic inference in subtribe Scandicinae. The trees obtained herein from analyses of morphological and anatomical data sets were not congruent to those obtained from separate analysis of ITS sequences and the relationships proposed were not as strongly supported. Moreover, the analysis of combined data produced trees that were practically identical with those inferred from molecular data alone. Although fruit characters failed to confirm the monophyly of subtribe Scandicinae, most of its included genera as well as “crown” Scandicinae were each identified as monophyletic. Moreover, the relationships within Athamanta, Scandix, Osmorhiza, and Kozlovia were identical and those within Anthriscus similar to those inferred from molecular data. Nearly all genera of Scandicinae are well delimited on the basis of fruit morphology and anatomy. The exception is Tinguarra for it cannot be readily distinguished from its sister Athamanta. Athamanta sicula was once regarded as a member of Tinguarra (Bentham 1867; Drude 1898; Knees 1996), while the type of Tinguarra, Tinguarra cervariifolia, was placed by de Candolle (1830) in Athamanta. Both a previous study using morphological data (Spalik and Downie 2001) and this study failed to provide clear diagnostic characters in which to distinguish Athamanta from Tinguarra. Since these two genera are undoubtedly closely related and morphologically indistinguishable, they have been combined (Spalik et al. 2001) and our study supports such a treatment. The position of Todaroa aurea is enigmatic. Our analysis of fruit characters places this species sister to the Athamanta–Tinguarra clade, while molecular data suggest an affinity that is not so straightforward. The isolated position of Todaroa in many trees is supported by its winged fruits, which are unique within the clade. The inclusion of Balansaea, exemplified here by Conopodium glaberrimum © 2001 NRC Canada 1370 Can. J. Bot Vol. 79, 2001 Fig. 4. The evolution of fruit morphology and anatomy in Scandiceae subtribe Scandicinae. Fruit sections are not to scale. Scale bar a is for the whole fruits of Myrrhis, Scandix, and Osmorhiza; scale bar b is for all remaining species. Drawings are by A. Wojewódzka. Anthriscus cerefolium Osmorhiza mexicana Anthriscus kotschyi Osmorhiza depauperata Anthriscus nitida Osmorhiza berteroi Anthriscus sylvestris Osmorhiza claytonii Ancestor of Kozlovia Ancestor of Osmorhiza Kozlovia paleacea Kozlovia longiloba Geocaryum macrocarpum Ancestor of Scandix and "Crown" Clade Kozlovia capnoides Ancestor of Scandix Scandix stellata Myrrhis odorata Scandix balansae Scandix pecten-veneris Scandix iberica a b 5 mm Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Ancestor of Anthriscus Chaerophyllum hirsutum Chaerophyllum temulum Chaerophyllum aromaticum Chaerophyllum bulbosum Chaerophyllum nodosum Chaerophyllum tainturieri Chaerophyllum aureum Chaerophyllum crinitum Chaerophyllum procumbens Chaerophyllum astrantiae Ancestor of Chaerophyllum Chaerophyllum macrospermum Sphallerocarpus gracilis Tinguarra montana Conopodium majus Ancestor of Basal Clade Conopodium glaberrimum Tinguarra cervariifolia Athamanta cretensis Athamanta turbith Todaroa aurea Common Ancestor of Scandicinae Athamanta sicula © 2001 NRC Canada Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Spalik et al. (; Balansaea glaberrima), in Conopodium, as advocated by Engstrand (1973) and adopted in this study, was confirmed by morphological data (Spalik and Downie 2001, and this study) but not by ITS data alone. The morphologically variable genus Chaerophyllum is well supported in molecular analyses (Downie et al. 2000a) but when morphology is considered the bootstrap support values are very low with one species, Chaerophyllum macrospermum, constantly grouped with Grammosciadium platycarpum. Interestingly, Koso-Poljansky (1916) placed Chaerophyllum macrospermum in the monotypic genus Golenkinianthe presumed to be related to Grammosciadium. This similarity is apparently homoplastic. We have demonstrated that although fruit morphology and anatomy appear inferior to molecular data for phylogenetic inference in subtribe Scandicinae, they are still more informative than other morphological characters. Spalik and Downie (2001) found that of the 44 discrete vegetative, inflorescence, floral, and fruit characters they examined in the subtribe, those relating to the fruit were most congruent with a phylogeny inferred from molecular data. RC values for these 44 characters ranged from 0 to 0.48 (the former referring to three characters that were completely homoplastic) and averaged 0.13. The six highest RC values, all greater than 0.25, pertained to characters of the fruit. In this study, RC values for 10 characters across a comparable array of taxa exceeded 0.25 (range 0.01–1.00, mean 0.20). When combined with molecular data, fruit characters did not change the tree topology but aided to resolve terminal branches where molecular characters alone were not variable enough. Fruit evolution in subtribe Scandicinae Based on the distribution of morphological character changes we infer that within Scandicinae, members of Athamanta have fruits that have retained many plesiomorphic character states and, thus, are likely very similar to those of the hypothetical common ancestor of the subtribe (Fig. 4). These fruits were probably oblong-ovate, somewhat laterally compressed (i.e., with mericarps as broad as wide), hairy, and with filiform ridges and constricted commissures. Its endosperm was probably deeply grooved, as seen in Smyrnium olusatrum (which is sister to the entire tribe) or as in some earlier branching Scandicinae (such as Sphallerocarpus gracilis or Chaerophyllum nodosum). The number of vallecular vittae was most likely variable (one to four), and these were irregular and anastomosing. The reconstruction of this state in the ancestor of the subtribe depends upon the choice of outgroup. Vittae are single in most Daucinae and Torilidinae, as well as in Todaroa, whereas early branching Daucinae (i.e., Laserpitium petrophilum), Smyrnium olusatrum, basal Scandicinae (Athamanta, Tinguarra, most members of Conopodium), and Sphallerocarpus have fruits with numerous vittae. Noteworthy is that in young fruits of Myrrhis odorata two types of vallecular vittae are seen: numerous small vittae situated close to the endosperm and somewhat larger ones, one per vallecula, that are placed outwards (Drude 1898). In Anthriscus, vallecular vittae are usually single, but sometimes, additional ones can occur (Spalik 1997). In young fruits of Anthriscus cerefolium the arrangement and number of vittae is similar to that occurring in Myrrhis (Kowal et al. 1969). 1371 Variation in fruit morphology and anatomy in the earliest branch of Scandicinae is minor; the number of vallecular vittae is reduced in Todaroa and some species of Conopodium, the indumentum is lost in Conopodium, and lateral wings occur in Todaroa. Fruits of these taxa are generally quite similar, with the differences among them being quantitative rather than qualitative. Many plesiomorphic characters are also retained in Sphallerocarpus. In Chaerophyllum, vallecular vittae are reduced to one per vallecula and, along with the vascular bundles, become enlarged, particularly in Chaerophyllum nodosum, presumed sister to all other members of the genus. However, in some species, particularly the American Chaerophyllum procumbens, these structures may have undergone substantial reduction, although they remain broader than those of the common ancestor of the subtribe. Fruits of most species of the genus are glabrous; exceptions include Chaerophyllum nodosum that has fruits covered with stiff setose hairs (bristles) and three other species that comprise variants with pubescent fruits. These variants are, however, rare. The most important evolutionary change that occurred in the common ancestor of Scandix and “crown” Scandicinae was the thickening of the exocarp cell walls. Irregular thickening of the external walls resulted in an aculeate cuticle, while the reinforcement of the transverse cell walls produced an areolate appearance to the epidermis. In some genera the entire cuticle became thickened. Members of the “crown” clade are also characterized by reduced vallecular vittae; although they have remained extant in Myrrhis and Geocaryum, in the former they are deprived of resin at fruit maturity, while in the latter they are relatively small leaving much of the endosperm girth unprotected. The common ancestor of the “crown” clade presumably had relatively narrow fruits that were distinctly pentagonal in transverse section. Such fruits are extant in Osmorhiza, Geocaryum, and Kozlovia capnoides. In Kozlovia longiloba, Kozlovia paleacea, and Anthriscus, the “overgrown” endosperm compresses the mesocarp and pushes the entire pericarp outwards; hence, the angular ridges have remained only at the top of the fruit. Therefore, mericarps of Kozlovia capnoides, when dissected one third above the fruit base, differ from those of Kozlovia paleacea and Kozlovia longiloba. However, when cut at two thirds above the fruit base, they look similar (Fig. 4). The broad, thickened bundles of some Scandix species, and particularly Scandix iberica, are superficially similar to those of Chaerophyllum. However, in Scandix stellata, sister to all other examined Scandix species, these bundles are very small and similar to those seen among species of the “crown” clade. Moreover, because of a relatively short beak (as compared to other species of Scandix) and a distinctly aculeate epidermis (which is not as pronounced in the other species), its fruits are reminiscent of members of the “crown” clade. These similarities are plesiomorphic. The evolution of fruit indumentum is obscure, as its occurrence in the subtribe is erratic, with pubescent and glabrous fruits often found in the same species (e.g., Chaerophyllum temulentum, Chaerophyllum procumbens, Chaerophyllum tainturieri, and Anthriscus cerefolium). The form of indumentum also varies considerably. Members of © 2001 NRC Canada Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. 1372 Athamanta, Tinguarra, and Todaroa have fruits covered with hairs that are similar to those occurring elsewhere on the plant. Similarly, the stiff, setose hairs occurring on the fruits of Chaerophyllum nodosum are also indistinguishable from those arising on its stems and leaves. In contrast, those bristles covering the fruits of members of the “crown” clade, which usually arise from small tubercles, appear distinct from the indumentum on the vegetative organs. Sometimes these bristles are secondarily reduced to short teeth, as in Kozlovia longiloba, or they break off easily in mature fruits, as in the bristled-fruit variety of Anthriscus cerefolium (and other species of Anthriscus not considered herein); hence, these fruits are described as tuberculate rather than bristled. Bristles that often break off at fruit maturity also occur in Myrrhis and Kozlovia capnoides. Conversely, they are persistent in Osmorhiza, while in Kozlovia paleacea they are so enlarged that Heywood (1982) placed this species alongside the spiny-fruited umbellifers in the tribe Caucalideae. This study demonstrates further that several umbellifer fruit modifications may not only have arisen independently in several lineages but are also variable at lower taxonomic levels. Many authors, including Drude (1898), Calestani (1905), and Cerceau-Larrival (1962), have already postulated independent origins for several fruit characters. These studies, however, were based on evolutionary scenarios inferred from morphological and anatomical data that are not in agreement with those relationships estimated from cladistic analyses of molecular data. Moreover, earlier workers usually refrained from interpreting these evolutionary changes in terms of adaptive shifts; such explanations are essential to our understanding of plant diversity. Adaptive significance of fruit characters Structural properties of the fruit result from a trade-off between its two primary functions: seed dispersal and seed protection. One may expect that fruit morphology and anatomy are subject to strong selective pressures, for it is clear that different fruit appendages reflect different dispersal strategies. Hooks and sticky hairs are usually interpreted as adaptations to epizoochory, while wings facilitate wind dispersal (Howe and Westley 1986). Strong pericarps or massive seed walls can protect seeds not only against seed eaters but also against microorganisms (fungi, bacteria) and harsh weather conditions. Grubb et al. (1998) have shown that plants with massive seed walls have significantly higher nitrogen concentrations in the embryo-cum-endosperm fraction than those with thinner walls. Moreover, other studies have suggested that even small qualitative and quantitative differences in fruit structure observed at the infraspecific level may reflect adaptive shifts to local environmental conditions. For example, in Heterosperma pinnatum (Asteraceae), the percentage of achenes having awns is correlated with closed vegetation, lower spring precipitation, and higher summer precipitation (Venable et al. 1998). Also, Hroudova et al. (1997) found that the achenes of two subspecies of Bolboschoenus maritimus (Cyperaceae) occurring in different habitats vary with respect to the development of aerenchymatic tissue that determines their buoyancy. Subspecies compactus occurring in temporarily flooded terrestrial habitats possesses more buoyant achenes, whereas Can. J. Bot Vol. 79, 2001 subspecies maritimus, occupying littoral habitats, has less buoyant achenes. Meyer (1997) showed that achene mass variation in the North American shrub Chrysothamnus nauseous (Asteraceae) was under strong genetic control that was correlated with fruiting time. Additionally, those Chrysothamnus subspecies with the heaviest achenes are restricted to specialized edaphic environments (dunes and badlands) or late seral montane riparian communities, while subspecies that are widely distributed and occur in early seral habitats have less heavy achenes. One may surmise that the differences in fruit structure observed in Scandiceae subtribe Scandicinae also represent similar adaptive shifts. The arrangement of vittae and vascular bundles suggests that these structures serve to protect the endosperm. The resin-filled vittae not only constitute mechanical protection but also contain active compounds that are toxic to insects (Berenbaum 1981). The broadening of vascular bundles occurs because of the development of sclerenchymatic tissue, providing a strong barrier against intruders. Phloem elements may be additionally shielded by intrajugal vittae. The presence in the epidermis of wax and a thickened cuticle constitutes both mechanical and chemical barriers that impede penetration by insects and fungi (Harborne 1993; Kerstiens 1996). The indumentum on a leaf serves as another defense mechanism (Gutschick 1999). In several species of Scandicinae, e.g., members of Athamanta and Tinguarra and Chaerophyllum nodosum, the indumentum of the fruit is similar to that occuring on the vegetative organs and may therefore have the same, presumably defensive, function. In Scandicinae, it is noteworthy that different forms of fruit and endosperm protection rarely occur together. The densely hispid fruits of Athamanta do not have the enlarged bundles and vittae as do the naked fruits of Chaerophyllum. Those members of the “crown” clade characterized by a thickened cuticle usually have reduced vittae and bundles, while the presence of an indumentum is erratic. Interestingly, in some species (e.g., Myrrhis odorata, Kozlovia capnoides, and Anthriscus) the bristles break off easily at fruit maturity thus appearing only useful during early stages of fruit development (i.e., when the cell walls of the exocarp are still thin). Although little is known about the chemical defenses of these species, several compounds with antiproliferative activity have been found in the fruits of Anthriscus sylvestris (Ikeda et al. 1998). In fact, such compounds may be specifically allocated to fruits. In Pastinaca sativa, for instance, toxic furanocoumarins have a much higher concentration in fruits than in other parts of the plant and constitute the first line of defense against the parsnip webworm, Depressaria pastinacella (Zangerl et al. 1997; Zangerl and Nitao 1998). The allocation of these chemicals to fruits is correlated with the endosperm mass but also depends on the paternal genotype; its variation seems to reflect competition among seed genotypes (Zangerl and Nitao 1998). Some members of the “crown” clade have fruits with bristles persistent at fruit maturity, seemingly to serve epizoochory. The long fruit beak of Scandix is usually covered with bristles and may serve to anchor the fruit on animal fur. However, Scandix australis and Scandix turgida (not considered herein) have central fruits with shorter beaks © 2001 NRC Canada Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. Spalik et al. that are deprived of bristles. These fruits are therefore less exposed to removal by animals. Other fruits have heteromorphic mericarps, where the inner mericarp is smooth and the outer one is bristled (Hedge and Lamond 1972; K. Spalik, personal observation). Heterocarpy also occurs in some members of subtribes Torilidinae and Daucinae. It has been suggested that the spiny fruits are adapted for long-distance dispersal, while the naked ones are to maintain the local population (Jury 1986). Persistent bristles also occur in those members of Osmorhiza that are characterized by a pedicel-like appendage. This appendage may function similarly to the beak of Scandix in facilitating dispersal. Fruits of Osmorhiza occidentalis that lack this appendage, and those of Osmorhiza mexicana ssp. bipatriata in which it is poorly developed, are glabrous (Lowry and Jones 1984). The role of this bristled appendage is best demonstrated by plant population studies. Gene flow within Osmorhiza claytonii, Cryptotaenia canadensis, and Sanicula odorata seems to reflect differences in their dispersal abilities. It is lowest in the glabrous-fruited Cryptotaenia canadensis and highest in Sanicula odorata (Williams and Guries 1994; Williams 1994). Both Sanicula odorata and Osmorhiza claytonii have bristled fruits; however, those of Sanicula odorata are lighter and therefore better adapted to epizoochory than those of Osmorhiza claytonii. Closely related taxa with bristled and glabrous fruits also occur in Anthriscus. Spalik (1996) speculated that the fruit morphology of European Anthriscus caucalis may have affected its dispersal abilities. The rare glabrous-fruited variety now occurs only in the northwestern Mediterranean, while the bristle-fruited variety is a common weed that has also reached Argentina, North America, India, and New Zealand (Spalik 1997). Several fruit characters promoting animal dispersal are also common to the Hawaiian and South American species of Sanicula and their closest North American relatives, with the adaptation to epizoochory having been lost multiple times in the western North American members of this genus (Vargas et al. 1999). In Scandiceae subtribe Scandicinae, fruit characters provide their greatest utility at the generic level, with different patterns of fruit morphological and anatomical evolution occurring within the subtribe. In contrast, we are unable to detect a single synapomorphy supporting the monophyly of the subtribe. It is surprising how little is known about the adaptive significance of these fruit modifications. The analysis of fruit character variation on the phylogenetic tree cannot answer the question whether they actually serve dispersal or defense of seed from predation but can provide us with valuable insight into the evolution of these characters. Acknowledgements The authors thank Jean-Pierre Reduron for material; the many herbaria cited in the text for loans of specimens; and Deborah S. Katz-Downie, Jean-Pierre Reduron, Ernie Small, and Ronald L. Hartman for comments on the manuscript. This work was supported by grants to K.S. from the Polish Committee for Scientific Research (KBN 6 P04C 02611) and to S.R.D. from the U.S. National Science Foundation (DEB-9407712). 1373 References Bentham, G. 1867. Umbelliferae. In Genera plantarum. Vol. 1. Edited by G. Bentham and J.D. Hooker. Reeve, London. pp. 859–931. Berenbaum, M.R. 1981. Evolution of specialization in insect– umbellifer associations. Annu. Rev. Entomol. 35: 319–343. Calestani, V. 1905. Contributo alla sistematica delle ombrellifere d’Europa. Webbia, 1: 89–280. Cerceau-Larrival, M.-T. 1962. Plantules et pollens d’Ombellifères. Mem. Mus. Natl. Hist. Nat. Ser. B Bot. 14: 1–166. Constance, L. 1971. History of the classification of Umbelliferae (Apiaceae). In The biology and chemistry of the Umbelliferae. Edited by V.H. Heywood. Academic Press, London. pp. 1–8. de Candolle, A.P. 1830. Umbelliferae. In Prodromus systematis naturalis regni vegetabilis. Vol. 4. Edited by A.P. de Candolle. Treüttel & Würtz, Paris. pp. 55–250. Downie, S.R., and Katz-Downie, D.S. 1996. A molecular phylogeny of Apiaceae subfamily Apioideae: evidence from nuclear ribosomal DNA internal transcribed spacer sequences. Am. J. Bot. 83: 234–251. Downie, S.R., Katz-Downie, D.S., and Spalik, K. 2000a. A phylogeny of Apiaceae tribe Scandiceae: evidence from nuclear ribosomal DNA internal transcribed spacer sequences. Am. J. Bot. 87: 76–95. Downie, S.R., Katz-Downie, D.S., and Watson, M.F. 2000b. A phylogeny of the flowering plant family Apiaceae based on chloroplast DNA rpl16 and rpoC1 intron sequences: towards a suprageneric classification of subfamily Apioideae. Am. J. Bot. 87: 273–292. Downie, S.R., Watson, M.F., Spalik, K., and Katz-Downie, D.S. 2000c. Molecular systematics of Old World Apioideae (Apiaceae): relationships among some members of tribe Peucedaneae sensu lato, the placement of several island-endemic species, and resolution within the apioid superclade. Can. J. Bot. 78: 506–528. Downie, S.R., Ramanath, S., Katz-Downie, D.S. and Llanas, E. 1998. Molecular systematics of Apiaceae subfamily Apioideae: phylogenetic analyses of nuclear ribosomal DNA internal transcribed spacer and plastid rpoC1 intron sequences. Am. J. Bot. 85: 563–591. Downie, S.R., Plunkett, G.M., Watson, M.F., Spalik, K., KatzDownie, D.S., Valiejo-Roman, C.M., Terentieva, E.I., Troitsky, A.V., Lee, B.-Y., Lahham, J., and El-Oqlah, A. 2001. Tribes and clades within Apiaceae subfamily Apioideae: the contribution of molecular data. Edinb. J. Bot. 58: 301–330. Drude, C.G.O. 1898. Umbelliferae. In Die natürlichen Pflanzenfamilien. Vol. 3(8). Edited by A. Engler and K. Prantl. Wilhelm Engelmann, Leipzig, Germany. pp. 63–250. Engstrand, L. 1973. Generic delimitation of Bunium, Conopodium and Geocaryum (Umbelliferae). Bot. Not. 126: 146–154. Engstrand, L. 1977. Biosystematics and taxonomy in Geocaryum Cosson (Umbelliferae). University of Lund, Lund, Sweden. Felsenstein, J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution, 39: 783–791. Gerlach, D. 1972. Zarys mikrotechniki botanicznej. PWRiL, Warsaw, Poland. Grubb, P.J., Metcalfe, D.J., Grubb, E.A.A., and Jones, G.D. 1998. Nitrogen-richness and protection of seeds in Australian tropical rainforest: a test of plant defense theory. Oikos, 82: 467–482. Gutschick, V.P. 1999. Biotic and abiotic consequences of differences in leaf structure. New Phytol. 143: 3–18. © 2001 NRC Canada Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. 1374 Harborne, J.B. 1993. Introduction to ecological biochemistry. Academic Press, London. Hedge, I.C., and Lamond, J.M. 1972. Scandix L. In Flora of Turkey and the East Aegean Islands. Vol. 4. Edited by P.H. Davis. Edinburgh University Press, Edinburgh. pp. 325–330. Heywood, V.H. 1971. Systematic survey of Old World Umbelliferae. In The biology and chemistry of the Umbelliferae. Edited by V.H. Heywood. Academic Press, London. pp. 31–42. Heywood, V.H. 1982. Multivariate taxonomic synthesis of the tribe Caucalideae. In Les Ombellifères: Contributions Pluridisciplinaires à la Systématique. Actes du 2ème Symposium International sur les Ombellifères. Edited by A.-M. CauwetMarc and J. Carbonnier. Braun-Brumfield, Ann Arbor, Mich. Monogr. Syst. Bot. Mo. Bot. Gard. No. 6. pp. 727–736. Higgins, D.G., Bleasby, A.J., and Fuchs, R. 1992. CLUSTAL V: improved software for multiple sequence alignment. CABIOS, 8: 189–191. Holmgren, P.K., Holmgren, N.H., and Barnett, L.C. 1990. Index herbariorum. New York Botanical Garden, New York. Howe, H.F., and Westley, L.C. 1986. Ecology of pollination and seed dispersal. In Plant ecology. Edited by M.J. Crawley. Blackwell, Oxford, U.K. pp. 185–215. Hroudova Z., Moravcova, L., and Zakravsky, P. 1997. Effect of anatomical structure on the buoyancy of achenes of two subspecies of Bolboschoenus maritimus. Folia Geobot. Phytotax. 32: 377–390. Ikeda, R., Nagao, T., Okabe, H., Nakano, Y., Matsunaga, H., Katano, M., and Mori, M. 1998. Antiproliferative constituents in Umbelliferae plants. IV. Constituents in the fruits of Anthriscus sylvestris Hoffm. Chem. Pharm. Bull. (Tokyo), 46: 875–878. Jury, S.L. 1986. Fruit and leaf variation in the African species of the Umbelliferae tribe Caucalideae. Symb. Bot. Upsal. 26: 181–188. Katz-Downie, D.S., Valiejo-Roman, C.M., Terentieva, E.I., Troitsky, A.V., Pimenov, M.G., Lee, B.-Y., and Downie, S.R. 1999. Towards a molecular phylogeny of Apiaceae subfamily Apioideae: additional information from nuclear ribosomal DNA ITS sequences. Plant Syst. Evol. 216: 167–195. Kerstiens, G. 1996. Signalling across the divide: a wider perspective of cuticular structure–function relationships. Trends Plant Sci. 1: 125–129. Kim, J.H. 1996. General inconsistency conditions for maximum parsimony: effects of branch lengths and increasing numbers of taxa. Syst. Biol. 45: 363–374. Kimura, M. 1980. A simple method for estimating evolutionary rates of base substitution through comparative studies of nucleotide sequences. J. Mol. Evol. 16: 111–120. Knees, S.G. 1996. Tinguarra Benth. & Hook. in Morocco. Lagascalia, 18: 286–287. Koch, W.D.J. 1824. Generum tribuumque plantarum Umbelliferarum nova dispositio. Nova Acta Acad. Caes. Leop. Carol. German. Nat. Cur. 12: 55–156. Koso-Poljansky, B. M. 1916. Sciadophytorum systematis lineamenta. Bull. Soc. Imp. Nat. Moscou, 29: 93–222. Kowal, T., Latowski, K., Macher, Z., and Pic., S. 1969. Anatomiczne cechy diagnostyczne leczniczych i trujacych gatunków Anthriscus Pers. Pr. Kom. Farm. Poznan. Tow. Przyj. Nauk, 7: 39–53. Lee, B.-Y., and Downie, S.R. 1999. A molecular phylogeny of Apiaceae tribe Caucalideae and related taxa: inferences based on ITS sequence data. Syst. Bot. 24: 461–479. Can. J. Bot Vol. 79, 2001 Lee, B.-Y., and Downie, S.R. 2000. Phylogenetic analysis of cpDNA restriction sites and rps16 intron sequences reveals relationships among Apiaceae tribes Caucalideae, Scandiceae and related taxa. Plant Syst. Evol. 221: 35–60. Lowry, P.P., and Jones, A.G. 1984. Systematics of Osmorhiza Raf. (Apiaceae: Apioideae). Ann. Mo. Bot. Gard. 71: 1128–1171. Maddison, W.P., and Maddison, D.R. 1992. MacClade. Analysis of phylogeny and character evolution, version 3 edition. Sinauer Associates, Sunderland, Mass. Meyer, S.E. 1997. Ecological correlates of achene mass variation in Chrysothamnus nauseosus (Asteraceae). Am. J. Bot. 84: 471–477. Morison, R. 1672. Plantarum umbelliferarum distributio nova. Sheldonian Theatre, Oxford, U.K. Plunkett, G.M., Soltis, D.E., and Soltis, P.S. 1996. Evolutionary patterns in Apiaceae: inferences based on matK sequence data. Syst. Bot. 21: 477–495. Rompel, J. 1895. Krystalle von Calcium oxalat in der Fruchtwandt der Umbelliferen und ihre Verwerthung für die Systematik. Sitzungsber. Akad. Wiss. Wien, 104: 417–476. Spalik, K. 1996. Species boundaries, phylogenetic relationships and ecological differentiation in Anthriscus (Apiaceae). Plant Syst. Evol. 199: 13–32. Spalik, K. 1997. Revision of Anthriscus (Apiaceae). Pol. Bot. Stud. 13: 1–69. Spalik, K., and Downie, S.R. 2001. The utility of morphological characters for inferring phylogeny in Scandiceae subtribe Scandicinae (Apiaceae). Ann. Mo. Bot. Gard. 88: 270–301. Spalik, K., Wojewódzka, A., and Downie, S.R. 2001. Delimitation of genera in Apiaceae with examples from Scandiceae subtribe Scandicinae. Edinb. J. Bot. 58: 331–346. Swofford, D.L. 1998. PAUP*. Phylogenetic analysis using parsimony (*and other methods), version 4 edition. Sinauer Associates, Sunderland, Mass. Theobald, W.L. 1971. Comparative anatomical and developmental studies in the Umbelliferae. In The biology and chemistry of the Umbelliferae. Edited by V.H. Heywood. Academic Press, London. pp. 177–197. Vargas, P., Baldwin, B.G., and Constance, L. 1999. A phylogenetic study of Sanicula sect. Sanicoria and S. sect. Sandwicenses (Apiaceae) based on nuclear rDNA and morphological data. Syst. Bot. 24: 228–248. Venable, D.L., Dyreson, E., Pinero, D., and Becerra, J.X. 1998. Seed morphometrics and adaptive geographic differentiation. Evolution, 52: 344–354. Williams, C.F. 1994. Genetic consequences of seed dispersal in three sympatric forest herbs: II. Microspatial genetic structure within populations. Evolution, 48: 1959–1972. Williams, C.F., and Guries, R.P. 1994. Genetic consequences of seed dispersal in three sympatric forest herbs. I. Hierarchical population-genetic structure. Evolution, 48: 791–805. Zangerl, A.R., and Nitao, J.K. 1998. Optimal defense, kin conflict and the distribution of furanocoumarins among offspring of wild parsnip. Evol. Ecol. 12: 443–457. Zangerl, A.R., Green, E.S., Lampman, R.L., and Berenbaum, M.R. 1997. Phenological changes in primary and secondary chemistry of reproductive parts in wild parsnip. Phytochemistry, 44: 825–831. © 2001 NRC Canada This article has been cited by: Can. J. Bot. Downloaded from www.nrcresearchpress.com by Canadian Science Publishing on 06/08/15 For personal use only. 1. Marcin Piwczyński, Radosław Puchałka, Krzysztof Spalik. 2015. The infrageneric taxonomy of Chaerophyllum (Apiaceae) revisited: new evidence from nuclear ribosomal DNA ITS sequences and fruit anatomy. Botanical Journal of the Linnean Society n/a-n/a. [CrossRef] 2. G.M. Plunkett, G.T. Chandler, P.P. Lowry, S.M. Pinney, T.S. Sprenkle, B.-E. van Wyk, P.M. Tilney. 2004. Recent advances in understanding Apiales and a revised classification. South African Journal of Botany 70, 371-381. [CrossRef] 3. Stephen R Downie, Ronald L Hartman, Feng-Jie Sun, Deborah S Katz-Downie. 2002. Polyphyly of the spring-parsleys (Cymopterus): molecular and morphological evidence suggests complex relationships among the perennial endemic genera of western North American Apiaceae. Canadian Journal of Botany 80:12, 1295-1324. [Abstract] [PDF] [PDF Plus]