MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System GALLEY 400 PHYLOGENY OF THE TRIBE HYMENOCALLIDEAE (AMARYLLIDACEAE) BASED ON MORPHOLOGY AND MOLECULAR CHARACTERS 1 mobt 89_303 Mp_400 File # 03em Alan W. Meerow, 2 Charles L. Guy, 3 Qin-Bao Li, 3 and Jason R. Clayton 4 ABSTRACT The generic limits of Hymenocallis have been variously proposed by different taxonomic workers, often without discussion or data. The genera Leptochiton, Ismene, Elisena, and Pseudostenomesson have been included with Hymenocallis, lumped together as the genus Ismene, or maintained as distinct genera. Recent cladistic analysis of plastid and nrDNA for Amaryllidaceae support a distinct tribe Hymenocallideae. Cladistic analyses of morphology, and plastid (trnL-F region) and nuclear ribosomal DNA (ITS) are presented alone and in combination for the tribe. Leptochiton is sister to the rest of the genera in the tribe in all analyses. While Hymenocallis is always resolved as monophyletic, Ismene is variably paraphyletic or monophyletic. The combined sequence data produce the most resolved and bestsupported phylogeny, wherein Hymenocallis and Ismene are monophyletic sister genera. These data support an origin for the tribe in the Andes, with vicariant distribution of the largely Mesoamerican Hymenocallis. Formal recognition of Ismene subg. Elisena and Pseudostenomesson is established. Key words: Amaryllidaceae, cladistics, molecular systematics, phylogeny. Systematics of the genus Hymenocallis Salisb. (Amaryllidaceae) and its allies have defied precise systematic understanding at both the specific and generic levels (Flory, 1976; Meerow & Dehgan, 1985). The genera Hymenocallis and Ismene Salisb. were established by Salisbury (1812) for the Neotropical species with fleshy seeds originally assigned to the Old World genus Pancratium L. The zygomorphic-flowered Elisena was described by Herbert (1837), who recognized Hymenocallis and Ismene as distinct genera. Baker (1888) subsumed Ismene within Hymenocallis but retained Elisena as distinct, as did Pax (1890). While Stapf (1933) treated H. quitoensis Herb. as a species of Pamianthe Stapf, Sealy (1937) considered the species to exhibit sufficient morphological divergence to be recognized as a monotypic genus, Leptochiton Sealy. Hutchinson (1934, 1959) retained both Elisena and Ismene (presumably including Leptochiton) as distinct. Velarde (1949) established the Peruvian genus Pseudostenomesson for a fleshy-seeded species originally described as Stenomesson morissonii Vargas as well as one new species. Traub (1962) recognized all four erstwhile genera as subgenera of Hymenocallis in his synoptic treatment: subg. Hy- menocallis, subg. Ismene (Salisb.) Baker ex Traub (including Leptochiton), subg. Elisena (Herb.) Traub, and subg. Pseudostenomesson (Velarde) Traub. Traub (1980) later reduced these subgenera to the rank of section without explanation. Ravenna (1980) in his description of H. heliantha (5 Leptochiton heliantha (Ravenna) Gereau & Meerow) suggested that subgenera Ismene (including Lepidochiton), Elisena, and Pseudostenomesson should probably be all recognized as the genus Ismene, distinct from Hymenocallis. Meerow and Dehgan (1985) suggested that Pseudostenomesson might warrant recognition at the rank of genus due to its extreme phenetic divergence (funnelform-tubular perianth) versus the ‘‘pancratioid’’ flower of Leptochiton, Ismene subg. Ismene, and Hymenocallis. ‘‘Pancratoid’’ floral morphology refers to a large, white, fragrant, crateriform flower with a conspicuous staminal cup (cf. Pancratium L.). This type of flower appears to be adapted for sphingid moth pollination (Bauml, 1979; Grant, 1983; Morton, 1965). Meerow (1990) treated Leptochiton as a distinct genus and recognized Hymenocallis and Ismene (including Elisena and Pseudostenomesson) as distinct, a treatment followed by Gereau et al. (1993) and This work was supported in part by NSF Grant DEB 9628787 to AWM and GLG. USDA-ARS-SHRS, 13601 Old Cutler Rd., Miami, Florida 33158, U.S.A., and Fairchild Tropical Garden, 11931 Old Cutler Road, Miami, Florida 33156, U.S.A. 3 University of Florida-IFAS, Department of Environmental Horticulture, 1545 Fifield Hall, Gainesville, Florida 32611, U.S.A. 4 USDA-ARS-SHRS, 13601 Old Cutler Rd., Miami, Florida 33158, U.S.A. 1 2 ANN. MISSOURI BOT. GARD. 89: 400–413. 2002. MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System GALLEY 401 mobt 89_303 Mp_401 File # 03em Volume 89, Number 3 2002 Meerow et al. Phylogeny of Tribe Hymenocallideae Meerow and Snijman (1998). No cladistic analysis has focused exclusively on testing the validity of this treatment, although at least one representative of each subgenus was included in overall molecular studies of Amaryllidaceae (Meerow et al., 1999, 2000a). Hymenocallis and its allied segregate genera are entirely Neotropical in distribution [a single West African taxon, H. senegambica, was treated by Sealy (1954) as an early adventive introduction of H. caribaea]. Hymenocallis sensus stricto, with 50 to 60 species, is chiefly Mesoamerican and extends into the West Indies and the southeastern United States. It is sparingly represented in northern South America. Leptochiton Sealy (2 spp.), Ismene (ca. 10 spp.), Elisena Herb. (2 or 3 spp.), and Pseudostenomesson Velarde (2 spp.) are all endemic to the Central Andean region of South America. Hymenocallis, Ismene, and Leptochiton are contrasted in Table 1. Hymenocallis and allies have usually been allied with Eucharis Planch. in the tribe Eucharideae (Hutchinson, 1934, 1959; Traub, 1963; Dahlgren et al., 1985; Müller-Doblies & Müller-Doblies, 1996). Meerow (1989, 1995) argued that the linkage of these genera, largely through the perception that both lineages shared a fleshy seed, was misconstrued, and proposed that either subtribal or tribal recognition of Hymenocallis and allies was warranted. Müller-Doblies and Müller-Doblies (1996) placed them in Eucharideae subtribe Hymenocallidinae, while Meerow and Snijman (1998) recognized a distinct tribe, Hymenocallideae. Family-wide analysis of plastid sequences (Meerow et al., 1999) and nrDNA analyses of the monophyletic American clade of the family (Meerow et al., 2000a) support a distinct Hymenocallideae as sister to the newly recognized tribe Clinantheae (a segregate of the former Stenomesseae), but complete resolution of the intratribal relationships is not apparent in these large analyses. Both tribes are subclades of a well-supported, Andean, tetraploid clade of genera. In this paper, we present phylogenetic analyses of morphological and molecular data for the tribe Hymenocallideae, and seek to clarify the relationships within the tribe. were used for one species each of the three subgenera of Ismene, one species of Leptochiton, and the outgroup Pamianthe peruviana (Table 2, Meerow et al., 1999). For ITS and the morphological data matrix, we increased our sampling with an additional four species of Hymenocallis and two additional species of Ismene subg. Ismene (Table 2). The aligned sequence matrices are available from the first author (miaam@ars-grin.gov). MATERIALS AND METHODS SAMPLING Sequences for the plastid trnL-F region were newly obtained for H. eucharidifolia, which, along with H. latifolia, was used as an exemplar taxon of Hymenocallis (Table 2). Previously cited sequences 401 MORPHOLOGICAL DATA Morphological and cytological character state data were derived from the following sources: Traub (1962, 1980), Sealy (1954), Flory (1976), Velarde (1949), Bauml (1979), Meerow and Dehgan (1985); from examination of living material in research collections at the USDA, Miami, Florida; field observations of Hymenocallis, Ismene, and Leptochiton species; and examination of herbarium material. The morphological matrix consists of 12 species representing 4 genera and 23 characters (Tables 3, 4). SEQUENCE DATA The trnL-F (trnL intron and spacer region between trnL and trnF) matrix consisted of 6 taxa and 906 base positions. The nrDNA ITS sequence matrix (ITS1, 5.8s intron, ITS2) consisted of 12 taxa and 636 bp. DNA EXTRACTION, AMPLIFICATION, AND SEQUENCING PROTOCOLS Genomic DNA was extracted from silica gel dried leaf tissue as described by Meerow et al. (2000a). The trnL-trnF region was amplified using the primers of Taberlet et al. (1991) as described by Meerow et al. (1999). Amplification of the ribosomal DNA ITS1/5.8S/ITS2 region was accomplished using flanking primers (18S, 26S) AB101 and AB102 (Douzery et al., 1999), and the original White et al. (1990) internal primers ITS2 and 3 to amplify the spacers along with the intervening 5.8S sequence, as described by Meerow et al. (2000a). Amplified products were purified using QIAquick (Qiagen, Valencia, California) columns, following manufacturer’s protocols. All polymerase chain reactions (PCR) were performed on an ABI 9700 (Applied Biosystems, Foster City, California). Cycle sequencing reactions were performed directly on purified PCR products on the ABI 9700, using standard dideoxy cycle protocols for sequencing with dye terminators on either an ABI 377 or ABI 310 automated sequencer (according to the Comparison of the genera and subgenera of Amaryllidaceae tribe Hymenocallideae. Genus or subgenus Number of species Distribution Elongate pseudostem Floral morphology Ovules per locule Phytomelan on seed coat Chromosome number Absent Pancratioid, actinomorphic, large, white or yellow, fragrant, sessile, (sub)erect; tube long; staminal cup large and striped green within; free filament short and incurved. 16–20 Present 2n 5 34 Ismene subg. Ismene 5–7 Central Andes at low to high elevations Present Pancratioid, actinomorphic large, white or yellow, fragrant, 2–10, subsessile to pedicellate, horizontal or declinate, tube 6 long, staminal cup large, striped green within; free filament short and incurved. 2–4 Absent 2n 5 23–86, 46, 104–110 Ismene subg. Elisena 2–4 Peru and Ecuador at mid to high elevations Present Zygomorphic, large, white, not fragrant, 2–10, subsessile, declinate; tube short; staminal cup 6 large, deflexed from the tube, free filaments long and declinate. 2 Absent 2n 5 46 Ismene subg. Pseudostenomesson 2 Peru, above 3000 m Present Funnelform-tubular, actinomorphic, 6 small, green, not fragrant, numerous, pedicellate, pendulous; tube long; staminal cup subcylindrical, free filament straight. 2 Absent 2n 5 46 Hymenocallis ca. 50 SE U.S., West Indies, Mesoamerica Absent Pancratioid, actinomorphic, large, white, fragrant, 1 to many, mostly sessile, erect; tube long; staminal cup large or small and variable in shape, not striped within; free filament long and straight. 2–10 Absent 2n 5 46, 40 most common but variable mobt 89_303 Mp_402 File # 03em SW Ecuador and NW Peru, at low elevations GALLEY 402 2 Annals of the Missouri Botanical Garden Leptochiton MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System 402 Table 1. MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System Volume 89, Number 3 2002 Table 2. Vouchers and new GenBank accession numbers for DNA sequences of Hymenocallideae. All vouchers are deposited at FTG unless otherwise indicated. GenBank accession no. or previous citation Origin trnL gene trnL-F spacer Mexico Mexico Florida, USA Mexico Trinidad — — Meerow et al. (1999) AF411078 — — — Meerow et al. (1999) AF411079 — Meerow Meerow Meerow Meerow Meerow Meerow 2452 Meerow 2441 Meerow 2306 Peru Peru Peru — — Meerow et al. (1999) — — Meerow et al. (1999) AF411080 Meerow et al. (2000) Meerow et al. (2000) Sagastegui 15454 Peru Meerow et al. (1999) Meerow et al. (1999) Meerow et al. (2000) Meerow 2308 Meerow 1116 Meerow 2304 Peru Ecuador Peru Meerow et al. (1999) Meerow et al. (1999) Meerow et al. (1999) Meerow et al. (1999) Meerow et al. (1999) Meerow et al. (1999) Meerow et al. (2000) Meerow et al. (2000) Meerow et al. (2000) Meerow Meerow Meerow Meerow Meerow 2424 2433 2438 2439 2440 ITS et et et et et al. al. al. al. al. (2000) (2000) (2000) (2000) (2000) 403 mobt 89_303 Mp_403 File # 03em Meerow et al. Phylogeny of Tribe Hymenocallideae Hymenocallis acutifolia (Herb.) Sweet H. glauca M. Roem. H. latifolia (Mill.) Roem. H. eucharidifolia Bak. H. tubiflora Salisb. Ismene subg. Ismene Ismene amancaes (Ruiz & Pav.) Herb. I. hawkesii (Vargas) Gereau & Meerow I. narcissiflora Jacq. Ismene subg. Elisena (Herb.) Meerow I. longipetala (Lindl.) Meerow Ismene subg. Pseudostenomesson (Velarde) Meerow I. vargasii (Velarde) Gereau & Meerow Leptochiton quitoensis (Herb.) Sealy Pamianthe peruviana Stapf Voucher GALLEY 403 Taxon MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System 404 Annals of the Missouri Botanical Garden Table 3. Characters and character states used in the cladistic analyses of Hymenocallideae based on morphology. Character 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. mobt 89_303 Mp_404 File # 03em GALLEY 404 Elongate pseudostem Flower number Flowers sessile/pedicellate Flower habit Tube length Tube habit Perianth morphology Perianth symmetry Flower color Fragrance Staminal cup shape Staminal cup striping Free filament Free filament Pollen grain size Pollen grain Exine reticulum Ovules per locule Seed per locule Phytomelan on testa Seed coat Seed shape Most common diploid chromosome number States and coding absent (0); present (1) 2–10 (0); solitary (1); .10 (2) sessile (0); pedicellate (1) erect (0); declinate/horizontal (1); pendent (2) shorter than tepals (0); longer than or equal to tepals (1) straight (0); curved (1) pancratioid (0); funnelform-tubular (1); 6 funnelform (2) actinomorphic (0); zygomorphic (1) white (0); yellow (1); green (2) present (0); absent (1) rotate or funnelform (0); cylindrical (1) present (0); absent (1) incurved (0); straight (1); declinate (2) longer than cup (0); shorter than cup (1) very large (0); large (1); medium (2) auriculate (0); not (1) coarse (0); medium (1) .20 (0); 16–20 (1); 2–10 (2); 2–4 (3) numerous (0); 2–5 (1); 1 (2) present (0); absent (1) not fleshy (0); fleshy (1) flat, winged (0); globose (1) 46 (0); 34 (1); 46, 40 (2) manufacturer’s protocols; Applied Biosystems, Foster City, California). SEQUENCE ALIGNMENTS Both sequence matrices were readily aligned manually using the program Sequencher (GeneCodes, Inc., Ann Arbor, Michigan) as few gaps needed to be inserted. CLADISTIC ANALYSES Pamianthe (tribe Clinantheae) was used as outgroup for all analyses. In larger sequence analyses (Meerow et al., 1999, 2000a), this genus resolves as most closely related to the Hymenocallideae. Pamianthe and Leptochiton (the latter putatively the least derived genus in the Hymenocallideae; see discussion below) share two four-base sequence elements in the trnL-F region (bp325–328, 821–824) Table 4. Character state matrix for cladistic analysis of 23 morphological characters in Hymenocallideae. Polymorphisms: 1 5 (0,1); * 5 (0,1,2). Taxon Character Hymenocallis acutifolia Hymenocallis eucharidifolia Hymenocallis glauca Hymenocallis latifolia Hymenocallis tubiflora Ismene amancaes Ismene hawkesii Ismene longipetala Ismene narcissiflora Ismene vargasii Leptochiton quitoensis Pamianthe peruviana Matrix 1 2 12345678901234567890123 00001000000110000221112 0*001000000130000221112 0*001000000130000221112 0*001000000130000221112 0*001000000130000221112 10111100100001000321110 10111100000001000321110 10110021011120211321110 10111100000001000321110 10120110211110211321110 01001100100001000110111 10111000000001110000000 MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System mobt 89_303 Mp_405 File # 03em GALLEY 405 Volume 89, Number 3 2002 Meerow et al. Phylogeny of Tribe Hymenocallideae 405 that are absent from the rest of the Hymenocallideae. Phylogenetic analyses were run using PAUP* version 4.0b8 beta (Swofford, 1998). An exhaustive search of all possible tree topologies was conducted for trnL-F. For ITS, the morphological, and all combined analyses, branch and bound searches were conducted. Support for internal nodes of the trees was determined with 5000 replicates of branch and bound bootstrapping (Felsenstein, 1985) and by calculation of Bremer (1988) decay indices (DI) using the program TreeRot (Sorenson, 1999). A branch and bound search was implemented for each constraint statement postulated by TreeRot. A bootstrap value of 50–64% was considered weak, 65–74% moderate, and 75–100% strong support. Combining independent character matrices, whether both molecular or molecular and morphological, very often increases the resolution of the ingroup and the bootstrap support of the internal nodes of the phylogenetic trees (Olmstead & Sweere, 1994; Chase et al., 1995; Yukawa et al., 1996; Rudall et al., 1998; Soltis et al., 1998; Meerow et al., 1999). Nonetheless, there is controversy about whether different data sets should be analyzed separately or together (De Queiroz et al., 1995; Huelsenbeck et al., 1996). Congruence of the independent matrices has generally been demonstrated before they are combined, but it has also been argued that incongruence should not be a predetermined factor against doing so (Dubuisson et al., 1998; Seelanan et al., 1997). Miyamoto and Fitch (1995) argued that data sets should always be analyzed independently, as underlying assumptions, constraints, or weighting strategies will vary from data set to data set. Kluge (1989) and Nixon and Carpenter (1996) argued that simultaneous analysis of multiple data sets better maximizes parsimony and allows secondary signals to appear from the combined data. Bull et al. (1993), Rodrigo et al. (1993), and De Queiroz (1993) advocated combining data only after a statistical test of congruence, what Huelsenbeck et al. (1996) called ‘‘conditional combination.’’ Before combining the data sets, we performed a partition homogeneity test (Farris et al., 1994, 1995) on the variously combined matrices, using a branch and bound search. trees, Hymenocallis is monophyletic (bootstrap 5 89%, DI 5 1), while Ismene is paraphyletic. Ismene longipetala (subg. Elisena) and I. vargasii (subg. Pseudostenomesson) are sisters in both trees. Leptochiton is sister to both Hymenocallis and Ismene in one tree (Fig. 1A). The 6 apomorphies at the ancestral node are an increase in pollen grain size, auriculate pollen grains, reduction in ovule number from more than 20 to 16 to 20; reduction in number of seeds per locule; and evolution of globose, fleshy seeds. Apomorphies for Hymenocallis (Fig. 1A) are the absence of an elongate pseudostem, predominantly sessile and erect flowers, and 2n 5 46, 40 chromosomes. Other than Hymenocallis, the only clade with strong bootstrap support is the sister relationship of Ismene subg. Elisena and subgenus Pseudostenomesson (100%, DI 5 6), based on 7 apomorphies: perigone tube length reduction, nonpancratioid floral morphology, loss of floral fragrance, cylindrical staminal cup, and smaller nonauriculate pollen grains with less coarse exine reticulum. If all of the characters are ordered as irreversible, a single tree is found of length 5 48, with CI 5 0.67 and RI 5 0.88 (Fig. 1B). There is moderate bootstrap support for a monophyletic Ismene (65%; DI 5 2; apomorphies: elongate pseudostem, pedicellate and declinate/horizontal flowers, and 2–4 ovules per locule). There is weak support for the sister relationship of Hymenocallis and Ismene (56%, DI 5 1; apomorphies: reduction in ovule and seed number, respectively; and the loss of phytomelan from the testa). Leptochiton is moderately supported as sister to both (65%, DI 5 1; apomorphies: reduction in ovule and seed number and the evolution of a fleshy seed). Ismene subg. Ismene has a 91% bootstrap and DI 5 2. Ismene subg. Elisena (I. longipetala) and subgenus Pseudostenomesson (I. vargasii) are again sister groups with 100% bootstrap and a DI 5 9. A monophyletic Hymenocallis receives 87% bootstrap support with a DI 5 4. Hymenocallis latifolia, H. glauca, and H. eucharidifolia form a monophyletic group with bootstrap support of 60 and DI 5 1. This same tree topology (Fig. 1B) is 40 steps long with CI 5 0.80 and RI 5 0.80 if a branch and bound search is run with the topology as a constraint with all characters unordered. RESULTS trnL-F MORPHOLOGICAL MATRIX PLASTID SEQUENCES With all characters unordered, two most parsimonious trees (Fig. 1A, one shown) were found of length 5 37, consistency index (CI) 5 0.86, and retention index (RI) 5 0.88. Sixteen of the 23 characters used were parsimony informative. In both Using trnL-F sequences, which provide 7 parsimony-informative base substitutions, three equally most parsimonious trees are found of length 5 82, CI 5 0.99, and RI 5 0.88 (Fig. 2, one tree shown). All three trees resolve a monophyletic Ismene with MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System 406 GALLEY 406 Annals of the Missouri Botanical Garden mobt 89_303 Mp_406 File # 03em Figure 1. Cladograms for Hymenocallideae based on morphological characters. —A. One of two most parsimonious trees found with all characters unordered. —B. Single most parsimonious tree found if all characters are ordered as irreversible. Numbers above branches are branch lengths. Bootstrap percentages and decay indices (italic) are in parentheses. Vertical lines and numbers below branches are apomorphies along that branch (see Table 3). MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System mobt 89_303 Mp_407 File # 03em GALLEY 407 Volume 89, Number 3 2002 Meerow et al. Phylogeny of Tribe Hymenocallideae 407 Figure 2. One of the three most parsimonious trees found by cladistic analysis of plastid trnL-F DNA sequences for the Hymenocallideae. Numbers above branches are branch lengths; numbers below branches are bootstrap percentages, followed by decay indices (italic). The large arrow indicates a node that collapses in the strict consensus of all three trees. 81% bootstrap support (DI 5 2), and Leptochiton as sister to the rest of the tribe but without support. A monophyletic Hymenocallis is resolved as sister to Ismene in one tree (Fig. 2), but Hymenocallis and Ismene form a clade in all three (73% bootstrap, DI 5 1). Ismene subg. Ismene (I. narcissiflora) and Elisena (I. longipetala) are resolved as sister groups in all three trees with a bootstrap of 70% (DI 5 1). ITS SEQUENCES ITS provides 50 parsimony-informative characters, and 9 trees of length 5 209, CI 5 0.73, and RI 5 0.77 were found (Fig. 3). In all of the trees, Leptochiton is resolved as sister to both Hymenocallis and Ismene (Fig. 3A), but without significant support. Hymenocallis is monophyletic (bootstrap 5 97%, DI 5 5), but Ismene is monophyletic in only 2 of the 9 trees (Fig. 3B, one shown). However, Ismene subg. Ismene (I. amancaes, I. hawkesii, I. narcissiflora) is monophyletic with weak bootstrap support (59%) and DI 5 1 (Fig. 3B). COMBINED trnL-F AND ITS SEQUENCES The P value from the partition homogeneity test 5 0.93, indicating that the trnL-F and ITS sequence matrices were highly congruent. Six most parsimonious trees were found of length 5 292, CI 5 0.92, and RI 5 0.77 (Fig. 4). In all trees, Hymenocallis and Ismene are monophyletic sister genera with bootstrap support of 94% and a DI 5 3. Leptochiton is sister to both, but without significant support. Bootstrap support for a monophyletic Hymenocallis is 98% (DI 5 5), but only 68% (DI 5 1) for a monophyletic Ismene. The only other inter- MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System 408 GALLEY 408 Annals of the Missouri Botanical Garden mobt 89_303 Mp_408 File # 03em Figure 3. Cladograms for Hymenocallideae based on nrDNA ITS sequences. —A. Strict consensus of nine equally parsimonious trees. —B. One of two trees in which Ismene is a monophyletic sister group to Hymenocallis. Numbers above branches are branch lengths; numbers below branches are bootstrap percentages followed by decay indices (italic). MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System GALLEY 409 Volume 89, Number 3 2002 mobt 89_303 Mp_409 File # 03em Meerow et al. Phylogeny of Tribe Hymenocallideae 409 Figure 4. One of six most parsimonious trees found by cladistic analysis of combined plastid trnL-F and nrDNA ITS sequences. Numbers above branches are branch lengths; numbers below branches are bootstrap percentages followed by decay indices (italic). The larger arrow indicates a node that collapses in the strict consensus of all six trees. nal resolution within Ismene that receives bootstrap support is a sister relationship between I. narcissiflora and I. hawkesii (both within subg. Ismene) at 84% with DI 5 2. COMBINED SEQUENCE AND MORPHOLOGICAL MATRICES The P value of the partition homogeneity test was 0.0003, indicating significant incongruence between the morphological and DNA sequence data matrices. Much of the apparent incongruence can be attributed to the weak resolution of the morphologically based topologies, and we felt that it would still be informative to combine the two matrices in a single analysis. Of the 1565 characters included, 76 were parsimony informative. A single tree was found of length 5 332, CI 5 0.92, and RI 5 0.79 (Fig. 5A). Hymenocallis is monophyletic with 100% bootstrap support (DI 5 8), but Ismene is paraphyletic. Bootstrap support for the monophyly of Ismene subg. Ismene rises to 81% (DI 5 2), but Ismene subg. Elisena (I. longipetala) and Pseudostenomesson (I. vargasii) are sister groups (bootstrap 5 97%, DI 5 4) weakly supported (bootstrap 5 57%, DI 5 1) as sister to Hymenocallis. Leptochiton is again sister to the other members of Hymenocallideae but without support. If trees one step longer were also retained in the search, in addition to the single shortest tree (Fig. 5A), a single, fully resolved tree of length 5 333, CI 5 0.90, and RI 5 0.77 was MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System 410 GALLEY 410 Annals of the Missouri Botanical Garden mobt 89_303 Mp_410 File # 03em Figure 5. Cladograms for Hymenocallideae based on combined morphological and DNA sequence matrices. —A. Single most parsimonious tree. Numbers above branches are branch lengths; numbers below branches are bootstrap percentages followed by decay indices (italic). —B. Single additional tree found if trees one step longer than tree pictured in 5A are retained. Numbers above branches are branch lengths. MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System GALLEY 411 mobt 89_303 Mp_411 File # 03em Volume 89, Number 3 2002 Meerow et al. Phylogeny of Tribe Hymenocallideae found (Fig. 5B). In this tree (Fig. 5B), both Hymenocallis and Ismene are monophyletic sister genera, as are Ismene subg. Elisena and Pseudostenomesson. Amaryllidaceae (Meerow et al., 2000a), with a vicariance event that gave rise to the largely North American Hymenocallis. Leptochiton, with 16 to 20 ovules per locule and a phytomelanous testa, occupies a relict position in the tribe with links to the non-fleshy seeded Andean endemic Clinantheae. However, it is the genus Ismene that reflects the patterns of floral morphological diversity that occur in the Eustephieae, Clinantheae, and Stenomesseae (sensu Meerow et al., 2000a). Ismene subg. Ismene retains the plesiomorphic pancratioid floral morphology of Leptochiton, Pamianthe, and Hymenocallis, while the smaller Ismene subg. Elisena and subg. Pseudostenomesson express floral novelties. Ismene subg. Pseudostenomesson, occurring at the highest elevations of any member of the tribe, might be the youngest element of the polymorphic Ismene, since the Andes likely did not extend above 1000 m elevation before the Pliocene (10 MYBP; Van der Hammen, 1974, 1979). Analogous patterns of floral diversity are found throughout the tetraploid Andean clade of the American Amaryllidaceae. In the Clinantheae, the low- to mid-elevation genera Pamianthe and Paramongaia Velarde have pancratioid floral morphology, while the mostly high-elevation Clinanthus Herb. has colorful, putatively ornithophilous flowers. In the more distantly related petiolate-leafed Stenomesseae, Eucharis has the pancratioid flower; Plagiolirion resembles a miniature Ismene subg. Elisena; and Stenomesson and Urceolina exhibit colorful, putatively ornithophilous flowers. Finally, in the Eustephieae, which is sister to rest of the Andean clade (Meerow et al., 2000a), the full range of variation is evident in a single genus, Hieronymiella Pax (Hunziker, 1969). This recurrent pattern suggests a scenario of rapid mosaic evolution (sensu Stebbins, 1984) within this monophyletic, tetraploid group (Meerow, 1987). The relatively low number of phylogenetically informative base substitutions in our sequence analyses of non-coding regions (7 for trnL-F; 50 for ITS) supports a hypothesis of a relatively recent radiation within the Hymenocallideae tied to the rise of the Andes. This seems most significant relative to Ismene, the most polymorphic of the three hymenocallid genera, and the only one that has adapted to high elevation. Hymenocallis is most speciose in Mexico (Bauml, 1979), with a secondary area of diversity in the southeastern United States (Smith & Flory, 1990, 2001; Smith et al., 2001). Only three described species have been reported from South America: the broadly and coastally distributed H. littoralis, H. pedalis, and H. tubiflora. The genus does not occur at all in the Andes, and H. tubiflora is the DISCUSSION Both plastid (Meerow et al., 1999) and ITS (Meerow et al., 2000a) sequences strongly support the position of the tribe Hymenocallideae as a monophyletic group within the Andean tetraploid clade of the endemic American Amaryllidaceae that is sister to the newly recognized tribe Clinantheae Meerow (Meerow et al., 2000a). The seeds of the Clinantheae are uniformly dry, flat, winged, and with phytomelanous testas. There are links between Leptochiton and Pamianthe that Stapf implicitly recognized, most notably the plesiomorphic presence of phytomelan in the testa of Leptochiton’s seed [of which Meerow & Dehgan (1985) were unaware], but also the numerous ovules of this genus (plesiomorphic as well). In the ITS phylogeny presented by Meerow et al. (2000a), support for Pamianthe as sister to the rest of Clinantheae (vs. a sister group relationship to Hymenocallideae or an unresolved position) was considerably weaker when the aligned matrix was not successively weighted. This is not surprising given that both genera occupy a basal phylogenetic position in their respective clades herein. The difficulty of relying on morphological characters alone to generate phylogenies in Amaryllidaceae has been discussed (Meerow, 1995; Meerow et al., 2000b), given a high degree of homoplasy for many morphological characters in the family. Our analysis (Fig. 1) generates trees with relatively high CI and RI, but parsimony is still not able to resolve Ismene nor consistently place Leptochiton in the basal position within the tribe with unordered morphological characters alone, in contrast to sequence data (Figs. 2–4), which also provide (in the combined trnL-F and ITS matrix), over three times the number of phylogenetically informative characters of morphology alone. The combined plastid and nuclear sequence matrix produces the most fully resolved shortest trees. To ‘‘force’’ this topology upon any of the other conflicting data matrices requires either ordering characters or accepting longer trees (albeit only one step longer in the combined sequence and morphological analysis). When biogeographic information is optimized upon the combined plastid and nrDNA tree (Fig. 4), the gene phylogeny supports an origin for the tribe in the central Andes, inarguably a locus of diversity for the Andean tetraploid clade of the 411 MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System GALLEY 412 mobt 89_303 Mp_412 File # 03em 412 Annals of the Missouri Botanical Garden only species of the three that is restricted to northern South America (including Trinidad-Tobago). The known distribution of the Hymenocallideae suggests two possible hypotheses, either a long-distance dispersal event from the Andean center of origin, or extinction of intervening populations of a proto-Hymenocallis ancestor. The fleshy seed of Hymenocallis is the largest of all the endemic American Amaryllidaceae, exhibits no dormancy, and germinates within 3–4 weeks after release, whether or not in substrate (Whitehead & Brown, 1940; pers. obs.). The relatively heavy seed does not immediately seem amenable to long-distance dispersal, and no dispersal agent other than water has even been suggested for the genus. Thus ancestral extinction is a more convincing hypothesis, but without a better understanding of the historical biogeography of Hymenocallis and a well-resolved phylogeny of the genus a likely explanation for its distribution cannot be determined. In summary, combined trnL-F and ITS sequences support the Meerow and Snijman (1998) treatment of Hymenocallideae with three genera: Hymenocallis, Ismene, and Leptochiton. Leptochiton is sister to the Hymenocallis/Ismene clade and retains two plesiomorphic characters of the Andean tetraploid clade: 16 to 20 ovules per locule and a phytomelanous seed coat. The central Andean endemism of Ismene and Leptochiton and the absence of Hymenocallis from this region further suggest a vicariance event at some point subsequent to the origin of the tribe. It is thus appropriate to formalize the recognition of the two new subgeneric combinations within Ismene. Bull, J. J., J. P. Huelsenbeck, C. W. Cunningham, D. L. Swofford & P. J. Waddell. 1993. Partitioning and combining data in phylogenetic analysis. Syst. Biol. 42: 384–397. Chase, M. W., D. W. Stevenson, P. Wilkin & P. J. Rudall. 1995. Monocot systematics: A combined analysis. Pp. 685–730 in P. J. Rudall, P. J. Cribb, D. F. Cutler & C. J. Humphries (editors), Monocotyledons: Systematics and Evolution, Vol. 2. Royal Botanic Gardens, Kew. Dahlgren, R. M. T., H. T. Clifford & P. F. Yeo. 1985. The Families of the Monocotyledons. Springer-Verlag, Berlin. De Queiroz, A. 1993. For consensus (sometimes). Syst. Biol. 42: 368–372. , M. J. Donoghue & J. Kim. 1995. Separate versus combined analysis of phylogenetic evidence. Ann. Rev. Ecol. Syst. 26: 657–681. Douzery, J. P., A. M. Pridgeon, P. Kores, H. Kurzweil, P. Linder & M. W. Chase. 1999. Molecular phylogenetics of Diseae (Orchidaceae): A contribution from nuclear ribosomal ITS sequences. Amer. J. Bot. 86: 887–899. Dubuisson, J. Y., R. Hebant-Mauri & J. Galtier. 1998. Molecules and morphology: Conflicts and congruence within the fern genus Trichomanes (Hymenophyllaceae) Molec. Phylogenet. Evol. 9: 390–397. Farris, J. S., A. G. Kluge & C. Bult. 1994. Testing significance of incongruence. Cladistics 10: 315–319. , M. Källersjö, A. G. Kluge & C. Bult. 1995. Constructing a significance test for incongruence. Syst. Biol. 44: 570–572. Felsenstein, J. 1985. Confidence limits on phylogenies: An approach using the bootstrap. Evolution 39: 783– 791. Flory, W. S. 1976. Distribution, chromosome number and types of various species of Hymenocallis. The Nucleus 19: 204–227. Gereau, R. E., A. W. Meerow & L. Brako. 1993. New Combinations in Hippeastrum, Ismene, and Leptochiton (Amaryllidaceae) for the Flora of Peru. Novon 3: 28– 30. Grant, V. 1983. The systematic and geographical distribution of hawkmoth flowers in the temperate North American flora. Bot. Gaz. 144: 439–449. Hammen, T. Van der. 1974. The Pleistocene changes of vegetation and climate in tropical South America. J. Biogeogr. 1: 2–26. . 1979. History of the flora, vegetation and climate in the Colombian Cordillera Oriental during the last five million years. Pp. 25–32 in H. Larsen & L. B. HolmNielsen (editors), Tropical Botany. Academic Press, London. Herbert, W. 1837. Amaryllidaceae. James Ridgeway and Sons, London. Huelsenbeck, J. P., J. J. Bull & C. W. Cunningham. 1996. Combining data in phylogenetic analysis. Trends Ecol. Evol. 11: 152–158. Hunziker, A. T. 1969. Estudios sobre Amaryllidaceae. III. Sinopsis provisional de Hieronymiella, y novedades Argentinas sobre Zephyranthes. Kurtziana 5: 343–367. Hutchinson, J. 1934. Families of Flowering Plants, Vol. 2. Monocotyledons, 1st ed. MacMillan, London. . 1959. Families of Flowering Plants, Vol. 2. Monocotyledons, 2nd ed. Clarendon Press, Oxford, U.K. Kluge, A. G. 1989. A concern for evidence and a phylogenetic hypothesis of relationships among Epicrates (Boidae, Serpentes). Syst. Zool. 38: 7–25. Meerow, A. W. 1987. The identities and systematic rela- Ismene subg. Elisena (Herbert) Meerow, comb. nov. Basionym: Elisena Herb., Amaryllidaceae, 75, 201. 1837. TYPE: Ismene ringens (Ruiz & Pav.) Gereau & Meerow, Novon 3: 29. 1993. Ismene subg. Pseudostenomesson (Velarde) Meerow, comb nov. Basionym: Pseudostenomesson Velarde. Rev. Cienc. (Lima) 51: 47–51. 1949. TYPE: Ismene vargasii (Velarde) Gereau & Meerow, in L. Brako & J. Zarucchi, Monogr. Syst. Bot. Missouri Bot. Gard. 45: 1253. 1993. Literature Cited Baker, J. G. 1888. Handbook of the Amaryllideae. George Bell and Sons, London. Bauml, J. A. 1979. A Study of the Genus Hymenocallis (Amaryllidaceae) in Mexico. M.S. Thesis, Cornell University, Ithaca, New York. Bremer, K. 1988. The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 198–213. MISSOURI BOTANICAL GARDEN Thursday Jul 25 2002 12:59 PM Allen Press • A V i i O N System GALLEY 413 mobt 89_303 Mp_413 File # 03em Volume 89, Number 3 2002 Meerow et al. Phylogeny of Tribe Hymenocallideae 413 tionships of Mathieua Klotzsch and Plagiolirion Baker (Amaryllidaceae). Taxon 36: 566–572. . 1989. Systematics of the Amazon lilies, Eucharis and Caliphruria (Amaryllidaceae). Ann. Missouri Bot. Gard. 76: 136–220. . 1990. 202, Amaryllidaceae. In G. Harling (editor), Flora of Ecuador No. 41. University of Göteborg. . 1995. Towards a phylogeny of the Amaryllidaceae. Pp. 169–179 in P. J. Rudall, P. J. Cribb, D. F. Cutler & C. J. Humphries (editors), Monocotyledons: Systematics and Evolution, Vol. 2. Royal Botanic Gardens, Kew. & B. Dehgan. 1985. The auriculate pollen grain of Hymenocallis quitoensis Herb. (Amaryllidaceae) and its systematic implications. Amer. J. Bot. 72: 540–547. & D. A. Snijman. 1998. Amaryllidaceae. Pp. 83– 110 in K. Kubitzki (editor), Families and Genera of Vascular Plants, Vol. 3. Springer-Verlag, Berlin. , M. F. Fay, C. L. Guy, Q-B. Li, F. Q. Zaman & M. W. Chase. 1999. Systematics of Amaryllidaceae based on cladistic analysis of plastid rbcL and trnL-F sequence data. Amer. J. Bot. 86: 1325–1345. , C. L. Guy, Q. Li & S-L. Yang. 2000a. Phylogeny of the American Amaryllidaceae based on nrDNA ITS sequences. Syst. Bot. 25: 708–726. , M. Fay, M. W. Chase, C. L. Guy, Q. Li, D. Snijman & S-L. Yang. 2000b. Phylogeny of the Amaryllidaceae: Molecules and morphology. Pp. 368–382 in K. Wilson & D. Wallace (editors), Monocots: Systematics and Evolution. CSIRO Press, Sydney. Miyamoto, M. M. & W. M. Fitch. 1995. Testing species phylogenies and phylogenetic methods with congruence. Syst. Biol. 44: 64–76. Morton, J. K. 1965. The experimental taxonomy of the West African species of Pancratium L. (Amaryllidaceae). Kew Bull. 19: 337–347. Müller-Doblies, D. & U. Müller-Doblies. 1996. Tribes and subtribes and some species combinations in Amaryllidaceae J. St.-Hil. emend. R. Dahlgren et al. 1985. Feddes Rep. 107(5–6): S.c. 1–9. Nixon, K. C. & J. M. Carpenter. 1996. On simultaneous analysis. Cladistics 12: 221–241. Olmstead, R. G. & J. A. Sweere. 1994. Combining data in phylogenetic systematics—An empirical approach using three molecular data sets in the Solanaceae. Syst. Biol. 43: 467–481. Pax, F. 1890. Amaryllidoideae–Narcisseae–Eucharidinae. In A. Engler & K. Prantl (editors), Die natürlichen Pflanzenfamilien II. Teil 5: 110–11. Wilhelm Engelmann, Leipzig. Ravenna, P. 1980. A new yellow-flowered Hymenocallis (Amaryllidaceae) from north Peru. Bot. Not. 133: 97– 98. Rodrigo, A. G., M. Kelly-Borges, P. R. Bergquist & P. L. Bergquist. 1993. A randomization test of the null hypothesis that two cladograms are sample estimates of a parametric phylogenetic tree. New Zealand J. Bot. 31: 257–268. Rudall, P. J., M. W. Chase, D. F. Cutler, J. R. Rusby & A. Y. De Bruijn. 1998. Anatomical and molecular systematics of Asteliaceae and Hypoxidaceae. Bot. J. Linn. Soc. 127: 1–42. Salisbury, R. A. 1812. On the cultivation of rare plants. Trans. Hort. Soc. London 1: 261–366. Sealy, J. R. 1937. Lepidochiton quitoensis. Bot. Mag. t. 9491. . 1954. Review of the genus Hymenocallis. Kew Bull. 2: 201–240. Seelanan, T., A. Schnabel & J. Wendel. 1997. Congruence and consensus in the cotton tribe (Malvaceae). Syst. Bot. 22: 259–290. Smith, G. L. & W. S. Flory. 1990. Studies on Hymenocallis henryae (Amaryllidaceae). Brittonia 42: 212–220. & . 2001. Hymenocallis (Amaryllidaceae) in Texas, with a new varietal combination. Novon 11: 229–232. , L. C. Anderson & W. S. Flory. 2001. A New Species of Hymenocallis (Amaryllidaceae) in the Lower Central Florida Panhandle. Novon 11: 233–240. Soltis, D. E., P. S. Soltis, M. E. Mort, M. W. Chase, V. Savolainen, S. B. Hoot & C. M. Morton. 1998. Inferring complex phylogenies using parsimony: An empirical approach using three large DNA data sets for angiosperms. Syst. Biol. 47: 32–42. Sorenson, M. D. 1999. TreeRot, version 2. University of Michigan, Boston University, Boston, Massachusetts. Stapf, O. 1933. Pamianthe peruviana. Bot. Mag. t. 9315. Stebbins, G. L. 1984. Mosaic evolution, mosaic selection and angiosperm phylogeny. Bot. J. Linn. Soc. 88: 149– 164. Swofford, D. L. 1998. PAUP: Phylogenetic Analysis Using Parsimony, v. 4.0 beta. Sinauer, Sutherland, Massachusetts. Taberlet, P., L. Gielly, G. Pautou & J. Bouvet. 1991. Universal primers for amplification of three non-coding regions of chloroplast DNA. Pl. Molec. Biol. 17: 1105– 1110 Traub, H. P. 1962. Key to the subgenera, alliances and species of Hymenocallis. Pl. Life 18: 55–72. . 1963. Genera of the Amaryllidaceae. American Plant Life Society, La Jolla, California. . 1980. Sections and alliances, genus Hymenocallis Salisb. Pl. Life 36: 46–49. Velarde, M. O. 1949. Pseudostenomesson, un nuevo genero de Amarilidáceas. Rev. Cienc. (Lima) 51: 47–51. White, T. J., T. Bruns, S. Lee & J. Taylor. 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. Pp. 315–322 in M. Innis, D. Gelfand, J. Sninsky & T. White (editors), PCR Protocols: A Guide to Methods and Applications. Academic Press, Orlando, Florida. Whitehead, M. R. & C. A. Brown. 1940. The seed of the spider lily, Hymenocallis occidentalis. Amer. J. Bot. 27: 129–198. Yukawa, T., H. Ohba, K. M. Cameron & M. W. Chase. 1996. Chloroplast DNA phylogeny of subtribe Dendrobiinae (Orchidaceae): Insights from a combined analysis based on rbcL sequences and restriction site variation. J. Pl. Res. 109: 169–176.