A revision of Passiflora L. subgenus Decaloba (DC.) Rchb. supersection Cieca (Medik.) J. M. MacDougal & Feuillet (Passifloraceae)

Abstract Passiflora subgenus Decaloba supersection Cieca is a monophyletic group of herbaceous to woody climbers found in subtropical and tropical regions of the world. The 19 species recognized here are primarily distributed in the southern United States, Mexico, Central America, South America, and the Caribbean. Two species, Passiflora suberosa and Passiflora pallida, are also naturalized in various regions of the Old World. The species of the supersection are recognized by their small, apetalous, usually greenish flowers with the filaments of the corona mostly in two series. The plants commonly lack c-glycosylflavones but possess flavonol 3-O-glycosides. The supersection contains two problematic species complexes, Passiflora suberosa and Passiflora coriacea. Phylogenetic relationships within supersection Cieca are investigated by means of phenetic and cladistic analyses of morphological and molecular (ITS 1 & 2) characters. The morphological and molecular data sets were analyzed separately because of incongruity due to taxon sampling and the complicated evolutionary history of entities within the Passiflora suberosa complex. All analyses confirm the monophyly of the supersection. They also show that the Passiflora suberosa complex is a non-monophyletic group of cryptic species, and inter-taxic hybridization and polyploidy have contributed to the confusing and complex pattern of variation evident within the group. Four taxa that were formerly included in this complex are recognized: Passiflora pallida, Passiflora suberosa subsp. suberosa, Passiflora suberosa subsp. litoralis, and Passiflora tridactylites. On the basis of molecular and morphological data, three species from the Passiflora coriacea complex are recognized: Passiflora coriacea, Passiflora sexocellata, and Passiflora megacoriacea. A key, detailed descriptions, distribution maps, and illustrations are included in the revision. Pollination, dispersal, and herbivory of the group are reviewed. The distribution and ecology of the species within the supersection are also discussed.


Introduction
Passifl ora L. subgenus Decaloba (DC.) Rchb. supersection Cieca (Medik.) J. M. Mac-Dougal & Feuillet is a monophyletic group of herbaceous to woody climbers found in subtropical and tropical regions of the world from latitude 34°N to latitude 34°S. Th e 19 species recognized here are primarily distributed in the southern United States, Mexico, Central America, South America, and the Caribbean. Two species, P. suberosa L. and P. pallida L., also occur in various regions of the Old World, likely as a result of naturalization.
Friedrich Medikus (Medikus 1787) was the fi rst to recognize this group, and he proposed the generic name Cieca for the apetalous species of Passifl oraceae. Since Medikus' time several monographers have also acknowledged the phenetic cohesiveness of these species and placed them (or most of them) in their own genus or section. However, the species with tubular, fl owers were often excluded and other species of uncertain relationship included. John MacDougal, as part of his revision of Passifl ora subgenus Decaloba section Pseudodysosmia (Harms) Killip [=supersection Bryonioides (Harms) J.M. MacDougal & Feuillet] (Passifl oraceae), was the fi rst to include the tubular-fl owered species in supersection Cieca and also to transfer various vegetatively divergent species out of the group (MacDougal 1983).
Five subgenera of Passifl ora are currently recognized: Passifl ora, Deidamioides (Harms) Killip, Astrophea (DC.) Mast., Decaloba (DC.) Rchb, and Tetrapathea (DC.) P.S. Green MacDougal 2003, Krosnick et al. 2009). Supersection Cieca belongs within subgenus Decaloba on the basis of having small (<4 cm in diameter) fl owers with the corona in a few series (two to three), and a plicate, membranous operculum. Th e base chromosome number of species in supersection Cieca and most species in subgenus Decaloba is six (n = 6); one count in a basal lineage of the subgenus is n = 9 (Snow and MacDougal 1993). Th e species of supersection Cieca are easily recognized by their small, apetalous, usually greenish fl owers with the fi laments of the corona mostly in two series. In addition, the fl owers lack bracts or possess only one or two bracts, and the plants commonly lack c-glycosylfl avones but possess fl avonol 3-O-glycosides.
Several factors enhance the biological signifi cance of Passifl ora supersection Cieca. Records of pollination are rare in the supersection, but the species exhibit three pollination syndromes: melittophily (pollination by bees), sphecophily (pollination by wasps), and ornithophily (pollination by birds) (Gilbert 1991;Koschnitzke and Sazima 1997;Lindberg 1998;MacDougal 1992). Th e species of the supersection are also utilized as larval hosts by most genera of the subfamily Heliconiinae (see section on herbivory) (Benson et al. 1975;Spencer 1988). Four of the 19 species within supersection Cieca are listed as endangered or threatened in the 1997 IUCN Red List of Th reatened Plants. One species, P. clypeophylla Mast., may be extinct and is represented by only a single herbarium specimen. Th e status of another species, P. macfadyenii C.D. Adams, is uncertain because, despite several searches, it has not been found in its native habitat in Jamaica since 1998. Supersection Cieca contains two problematic species complexes, P. suberosa and P. coriacea Juss. Ever since Linnaeus fi rst described P. suberosa in his Species Plantarum, taxonomists have disagreed about the circumscription of this widespread species and, as a result, many synonyms exist for it (Linnaeus 1753; Th e Herbarium of the Royal Botanic Gardens 1996). My analysis of the herbarium specimens of P. suberosa s. l. indicate that this variable species has served as a "disposal depot" for at least four entities (P. pallida, P. suberosa L. subsp. suberosa L., P. suberosa L. subsp. litoralis (Kunth) K.Port.-Utl. ex. M.A.M.Azevedo, Baumgratz & Gonç.-Estev., and P. tridactylites Hook.f.) that cannot be assigned to any of the other members of the supersection. Molecular and morphological phylogenetic analyses show that the complex is a non-monophyletic group of cryptic species, a situation not unusual in plants (Rieseberg and Brouillet 1994). Passifl ora coriacea Juss., as traditionally circumscribed, is another "species" that exhibits marked morphological variation over its distribution from eastern Mexico to northern South America, and evidence presented in this study indicates that it comprises three distinct entities (P. coriacea, P. megacoriacea K.Port.-Utl., and P. sexocellata Schltdl.).

Taxonomic history
Th e genus Passifl ora (Passifl oraceae Juss. ex Roussel; tribe Passifl oreae DC.) is a large and diverse group of approximately 500 species of vines, lianas, and trees (Feuillet and MacDougal 1999;Feuillet and MacDougal 2003;Killip 1938;Wilde 1971). Th e geographical distribution of Passifl ora is primarily restricted to New World tropical, subtropical, and occasionally temperate areas, but approximately 20 species are found in Southeast Asia, Oceania, and Australia. Th e genus Passifl ora contains fi ve subgenera: Passifl ora, Deidamioides (Harms) Killip, Astrophea (DC.) Mast., and Decaloba (DC.) Rchb., and Tetrapathea (DC.) P.S.Green MacDougal 2003, Krosnick et al. 2009). Th e two largest subgenera in the genus are Passifl ora (~250 species) and Decaloba (~230 species). Supersection Cieca, one of eight supersections in subgenus Decaloba, is the fourth largest supersection in the subgenus . Nineteen species are recognized in the supersection, and of those, two are newly described in this revision. Traditionally, the tubular-fl owered members of the group have been separated from those that possess dish-shaped fl owers; the tubular-fl owered species often have been placed in segregate sections and genera.
From 1570-1577 Francisco Hernández, the personal physician of King Philip II of Spain, traveled in the Americas in search of new medicines. Hernández spent his time in Mexico and enlisted native guides, artists, herbalists, and physicians to teach him about the materia medica, resulting in the earliest treatment of Mexico's natural history. However, it was not until 1651 that his manuscript was published in Rerum medicarum Novae Hispaniae thesaurus seu plantarum animalium mineralium Mexicanorum historia. In it was the fi rst description of a plant from supersection Cieca, Passifl ora sexocellata Schltdl. Hernández gave the Aztec name for it, Tzinacanatlapatli, followed by a brief description and illustration of the plant (Hernández 1651).
In the year 1719, Joseph Pitton de Tournefort created two genera of passionfl owers: Granadilla and Murucuia. One species with fused coronal fi laments was placed in the genus Murucuia (= P. murucuja L.). Th e remaining 23 species recognized by Tournefort, including the species of supersection Cieca described and illustrated by Plumier, were placed in the genus Granadilla (Tournefort 1719).
Th e next published monograph was Decima dissertatio botanica de Passifl ora (Cavanilles 1790) in which a total of 43 species (all placed in the genus Passifl ora) were described; 32 of the species were illustrated. One new species of supersection Cieca [P. peltata Cav. (= P. suberosa in this revision)] was included, in addition to the four described by Linnaeus. In 1799, Antonio Cavanilles, in his Icones et Descriptiones Plantarum, described another species of supersection Cieca, the tubular-fl owered P. viridifl ora Cav. (Cavanilles 1799).
In 1805, Antoine Laurent de Jussieu formally described 13 new species of the genus, including Passifl ora coriacea; he also recognized the genera Murucuia and Tacsonia. Because of its tubular fl owers, he placed the apetalous P. viridifl ora in the genus Tacsonia. In this treatment, Jussieu also discussed in great detail questions of generic delimitation and relationship, and he was the fi rst to suggest that Passifl ora, Murucuia, and Tacsonia should be placed together in their own family. However, Jussieu did not use a family name in the offi cial sense (Jussieu 1805a(Jussieu , 1805b. Henri Francois Anne de Roussel, in 1806, was the fi rst to validly publish the family Passifl oraceae, with credit to Jussieu. Until recently, the fi rst valid publication of the Passifl oraceae was attributed to Karl Sigismund Kunth; however, Roussel's description of the family was published more than ten years before Kunth's. Incidentally, Kunth in H.B.K. also published the species P. tubifl ora but was probably unaware that Cavanilles had already named it P. viridifl ora (Kunth 1817). In 1826, Curt Polycarp Joachim Sprengel transferred P. viridifl ora into the genus Murucuia (Sprengel 1826). One year later, William Hamilton, in his Prodromus Plantarum Indiae Occidentalis, described three new species of Passifl ora. One of these, P. lancifolia Ham., is a red, apetalous, hummingbird-pollinated member of supersection Cieca from the Antilles (Hamilton 1825).
In 1822 and 1828, Augustin Pyramus de Candolle subdivided Passifl ora into eight sections based upon bract and calyx morphology: Astrophea, Polyanthea, Tetrapathea, Cieca, Decaloba, Granadilla, Tacsonioides, and Dysosmia; he thought that all members of the tribe Passifl oreae lacked a corolla. He placed all of the species with dish-shaped fl owers that were either ebracteate or possessed small bracts and a fi ve-lobed calyx in section Cieca. However, he mistakenly placed individuals that we now know possess fi ve petals and fi ve sepals (a ten-lobed calyx, according to de Candolle) in the section. He did not place P. lancifolia in a section because he felt that the species was not suffi ciently known. Additionally, he placed Passifl ora viridifl ora in the section Psilanthus and accepted Jussieu's placement of the species in the genus Tacsonia (Candolle 1828). In the same year, Heinrich Gottlieb Ludwig Reichenbach (1828) raised section Decaloba DC. to the rank of subgenus. Additionally, he placed Passifl ora viridifl ora in the genus Synactila Raf.
Max Joseph Roemer (1846) published a monograph of the Passifl oraceae and raised de Candolle's sections to the rank of genera. Th us, most of the apetalous species discussed above were once again placed in their own genus, though Roemer repeated de Candolle's mistake and also placed petalous species in the genus Cieca. He placed P. lancifolia in the genus Decaloba and recognized Psilanthus viridifl ora (Roemer 1846). Soon after the publication of Roemer's monograph, Joseph Dalton Hooker supported de Candolle's broad and more conservative concept of the genus in his treatment for Genera plantarum (Hooker 1867). In the interim, another apetalous species belonging to supersection Cieca, P. tenuiloba Engelm., was described by George Engelmann (Gray 1850).
In 1871, Maxwell Tylden Masters published a preliminary taxonomic paper on the Passifl oraceae in the Transactions of the Linnaean Society that would be expanded upon in a comprehensive monograph of the family that appeared a year later in Carolus Martius' Flora Brasiliensis (Masters 1871(Masters , 1872. In the 1871 publication, he validly established four subgenera within Passifl ora based upon various fl oral characteristics: Astrophea (Ohwi) Rchb., Plectostemma Mast. (with sects. Cieca, Dysosmia, and Decaloba), Murucuia Tourn. ex Mill. (with sects. Eumurucuia and Psilanthus), and Granadilla Mill. Section Cieca was put in subgenus Plectostemma and consisted of the apetalous members of Passifl ora that lacked bracts and possessed dish-shaped fl owers; he also mistakenly placed some petalous species in the group. In addition, many of the species that previous authors recognized as distinct from P. suberosa were reduced to varieties (see discussions of P. pallida and P. suberosa for details). Masters also placed Passifl ora lancifolia and Passifl ora viridifl ora in an unnamed section, along with other tubular-fl owered species. In the 1872 monograph, he maintained section Cieca as described above, but he recognized not only varieties of P. suberosa but also subvarieties. In addition, Masters placed P. lancifolia and P. viridifl ora (and the associated species from the 1871 paper) in section Psilanthus Hook.f. He also (Masters 1871) put all of these in his subg. Plectostemma and appeared to be unaware that Reichenbach (1828) had already elevated Decaloba to the rank of subgenus. John Mochrie MacDougal (1983) pointed out that the type species of these two subgenera, subg. Plectostemma Mast. and subg. Decaloba (DC.) Rchb., are so closely related that for all practical purposes Decaloba shoud be used instead of Plectostemma. Jose JeronimoTriana and Jules Émile Planchon (1873), in their monograph of the Colombian Passifl oraceae adopted Masters' 1872 classifi cation with one modifi cation. Th ey reduced the genus Tacsonia to a subgenus within Passifl ora. In 1887 and 1891, Masters described two additional species of supersection Cieca that are endemic to Guatemala, P. trinifolia Mast. and P. clypeophylla Mast. (Masters 1887(Masters , 1891. In 1890, Martin Sessé y Lacasta and José Mariano Mociño, in Plantae Novae Hispaniae, post humously described P. obtusifolia Sessé & Moc. (here placed in supersection Cieca). Th ere is also an illustration of that species in Icones Florae Mexicanae (McVaugh 1977(McVaugh , 1980(McVaugh , 1982Sessé and Mociño 1887-1890, 1894. However, no later author until Ellsworth Killip (1938) mentioned P. obtusifolia. Hermann Harms, in his Die Natürlichen Pfl anzenfamilien (1893, 1897, 1925, revised the generic and infrageneric classifi cation of the family. Harms, instead of dividing the genus Passifl ora into subgenera, separated it into 21 sections; the sections were often divided into subsections or series. He also recognized the New World genera Dilkea, Mitostemma, and Tetrastylis. He defi ned the members of section Cieca as possessing small, whitish or greenish, bowl-shaped fl owers without petals. He mistakenly thought that P. inamoena A.Gray (= P. bryonioides Kunth) lacked petals and therefore included it in section Cieca. He placed the apetalous P. gracilis J.F.Jacq. ex Link in the section. However, MacDougal (1994) determined (based upon morphological evidence) that it is more closely related to species in supersection Bryonioides (Harms) Feuillet & MacDougal than to members of supersection Cieca. Harms did not indicate where P. lancifolia belonged, but he placed P. viridifl ora by itself in section Chloropathanthus Harms (Harms 1893(Harms , 1897(Harms , 1925. In 1938, Killip published a revision of the American Passifl oraceae. Killip's revision, by his own admission, closely approximated that of Harms. Th e most important diff erences were the raising of Harms' sections to subgenera and the regrouping of the species placed by Harms in sections Decaloba and Cieca. Killip defi ned section Cieca as those members of subgenus Plectostemma (= subgenus Decaloba) that possess peti-olar glands, reticulate seed coats, and bracts that are scattered along the peduncle and more than 1 mm long; he considered the lack of bracts in many of the species of the section to be the result of deciduousness. He placed in the group the apetalous species of Cieca Medik., the "Bryonioideae" of Harms, and several other species of uncertain relationship. However, in comparison with other genera in the Passifl oraceae, the character states that he used to defi ne the section are plesiomorphic, and the members of his section Cieca are now considered an artifi cial assemblage (MacDougal 1994). He placed P. viridifl ora and P. lancifolia in subgenus Chloropathanthus Harms, based upon the lack of a plicate operculum, even though all of the other characters that he used to defi ne this subgenus are the same as those that he used to designate section Cieca.

Morphological data set
Th e morphological investigation of supersection Cieca is based upon the careful study of over 4,200 dried specimens from 44 herbaria, supplemented with observations from plants preserved in ethyl alcohol and living plants in the fi eld and greenhouse. Of the 19 species in the supersection, 13 (all except P. clypeophylla, P. eglandulosa, P. macfadyenii, P. trinifolia, P. megacoriacea, and P. tridactylites) were collected during fi eld work in Jamaica, Haiti, and Mexico, or donated to the Passifl ora greenhouse collection at the University of Florida; several correspondents and colleagues contributed living material during the course of this study. Vouchers are deposited at CICY and FLAS.
An average of 330 macromorphological characters were measured or observed on each of 95 plant specimens. All of the herbarium specimens representing supersection Cieca were carefully observed, and those spanning the morphological variation and geographical range of each species were chosen for measurement. Depending upon the material available, up to fi ve measurements were taken for each quantitative character on each specimen. Characters were measured or scored from corresponding positions  on mature, reproductive plants from throughout the geographical range of the supersection in order to minimize error due to developmental diff erences. Measurements of the dried leaves of the species in supersection Cieca were taken according to the conventions in . Th e fl owers were measured in accordance with the standards in Figs 4-6. Dried fl owers from herbarium specimens were rehydrated by placing them in warm water with a wetting agent (Aerosol OT) or immersing them in concentrated ammonia (Toscano de Brito 1996;Taylor 1975). Color names used in this treatment follow the Munsell Color System (Long and Luke 2001). All drawings of fl owers were made either from material fi xed in standard FAA [70% ethyl alcohol (90%), glacial acetic acid (5%) and formalin (5%)] and preserved in 70% ethyl alcohol or from herbarium material that was expanded and softened.
Seventy quantitative characters were initially evaluated for the Neighbor Joining analysis of the P. suberosa complex, but 25 were discarded due to lack of variability or lack of unambiguous gaps in the pattern of variation, making state delimitations diffi cult. Of the 45 remaining, six show no overlap in the range of variation of their states. Th e remaining 39 quantitative characters were utilized even though they exhibit some arbitrariness in state delimitation. Seventy quantitative characters were initially evaluated for the Neighbor Joining analysis of the P. coriacea complex, but 37 were discarded due to lack of variability or lack of gaps in the pattern of variation, making state delimitations diffi cult. Of the 33 remaining, 17 show no overlap in the range of variation of their states. Th e remaining 16 quantitative characters were utilized even though they exhibit some overlap between delimited states.
Seventy quantitative characters were initially evaluated for the cladistic analysis of the supersection, but 31 were discarded due to lack of variability or problems in delimiting character states. Of the 39 remaining, only one shows no overlap in the range of variation of its states (see for example Fig. 7). Th e remaining 38 quantitative characters were utilized even though they exhibit some overlap in the range of variation assigned to diff erent character states between taxa (see for example Fig. 8). In all species descriptions, the fl ower diameters were mathematically determined: 2(sepal length) + hypanthium diameter.
Distribution maps were produced in ESRI® ArcMap TM 10.0 (Environmental Systems Research Institute, Inc., Redlands, California, USA). Label data from herbarium specimens were used to determine the latitudes and longitudes employed in the con-     struction of distribution maps. Th e gazetteer consulted for localities in the United States was the Geographic Names Information System (GNIS), developed by the U.S. Geological Survey (USGS) in cooperation with the U.S. Board on Geographic Names (US BGN). For international localities, the primary gazetteer consulted was the Geographic Names Database (http://www.nima.mil/geonames/GNS/index.jsp). Th e Geographic Names Database is on the GEOnet Names Server (GNS), the offi cial repository of foreign place-name decisions approved by the United States Board on Geographic Names (US BGN). A secondary source for international localities was the Tageo Database of Geographic Coordinate Information. Th e coordinate system for data served by the GNIS, GNS and the Tageo Database of Geographic Coordinate Information is WGS84. Principal components analyses (PCA) were produced using the computer program Multi-Variate Statistical Package (MVSP) 3.13d (Kovach Computing Services, Anglesey, Wales, UK). All measured quantitative characters for specimens in the P. suberosa and P. coriacea complexes were used in the analyses. Only those characters and specimens for which there was an abundance of missing data were deleted from the analyses. Th e mean values of measurements were used for each specimen. Box plots were produced using the computer program SPSS for Windows Release 11.5 (SPSS,Inc.,Chicago,Illinois,USA). Neighbor joining trees were produced in PAUP* Version 4.0b10 for Macintosh (Sinaeur, Sunderland, Massachusetts, USA). Morphological characters were also analyzed cladistically, and cladistic methods are discussed under the heading "Phylogenetic Search Strategies".

Molecular data set
Total genomic DNA was extracted from fresh, heat dried, or silica dried leaves or fl owers utilizing the CTAB method of Doyle and Doyle, scaled down to 1.0 ml extraction volumes (Doyle and Doyle 1987). Amplifi cation of the internal transcribed spacer (ITS) region of 18S-26S nuclear ribosomal DNA (nrDNA) was performed using 50 μl reactions, 2.5 mmol/L MgCl 2 , 1.0 mol/L betaine, and a hot start at 94 °C for between 3-10 minutes, using Epicentre (Epicentre Technologies, Madison, Wisconsin, USA) buff ers and Taq polymerase. For material extracted from herbarium specimens, the above amplifi cation protocol was modifi ed by using 3.0 mmol/L MgCl 2 , no betaine, and a hot start at 99 °C for 30 seconds. A touchdown thermal cycling program was used for fresh and silica-dried samples. An initial denaturation at 94 °C for three minutes was followed by an initial annealing temperature of 76 °C (decreasing 1 °C per cycle for 17 cycles to 59 °C), extension at 72 °C for one minute and denaturation at 94 °C for one minute. Th is was followed by 22 cycles of annealing at 59 °C for one minute, extension at 72 °C for one minute, denaturation at 94 °C for one minute and a fi nal extension at 72 °C for four minutes. Amplifi cation and sequencing primers were those of Sun et al. (1994). For the herbarium material, an initial denaturation at 94 °C for two minutes was followed by ten cycles of annealing at 55 °C for 30 seconds, extension at 72 °C for 30 seconds and denaturation at 94 °C for 20 seconds. Th is was followed by 33 cycles of annealing at 55 °C, extension at 72 °C for one minute, denaturation at 94 °C for 20 seconds and a fi nal extension at 72 °C for seven minutes. Amplifi cation and sequencing primers were those of Blattner (1999). PCR products were cleaned using QIAquick columns (Qiagen, Santa Clarita, California, USA) and underwent dye terminator cycle sequencing with Applied Biosystems Inc. (ABI) (Foster City, California, USA) reagents (5 μL reactions). Th e ITS region for members of supersection Cieca was sequenced directly from the cleaned amplifi ed product in the DNA Sequencing Core Facility on the University of Florida campus (DSEQ, UF) with ABI 377 and ABI 373A automated sequencers. Sequences were edited and assembled using the ABI software packages Sequence Navigator ™ Version 1.0.1 and Auto Assembler ™ Version 1.3.0 on an Apple PowerMac computer and aligned visually.
Selected cleaned PCR products were cloned into TOPO-TA Cloning® (Invitrogen, Carlsbad, California, USA) vectors according to the manufacturer's instructions, except that the ligation reactions were halved. Transformation reactions were incubated in SOC broth (2.0% tryptone, 0.5% yeast extract, 10 mM NaCl, 2.5 mM KCl, 10 mM MgCl 2 -6H 2 O, 20 mM glucose) at 37 °C for one hour before being spread onto plates containing S-Gal™/LB Agar/Kanamycin Blend (Sigma, St. Louis, Missouri, USA) and incubated at 37 °C for 8-18 hours. Only large white colonies, representing potentially recombinant plasmids, were selected for amplifi cation and sequencing.

Phylogenetic search strategies
Two matrices were analyzed cladistically for this study: morphology (32 taxa including outgroups) and ITS sequence data (71 taxa including outgroups). Th e morphological character states were carefully delimited (see discussion under "Morphological Data Set"). Many characters are qualitative, and discrete states were delimited within quantitative characters by assessment of gaps in the pattern of variation (Stevens 1991). Multistate characters were considered to be unordered and ingroup/outgroup relationships were analyzed simultaneously. Outgroups from the subgenera Decaloba (supersections Auriculata, Bryonioides, Decaloba, Hahniopathanthus, Multifl ora, and Pterosperma) and Deidamioides (section Tryphostemmatoides) were selected based on studies by two Passifl ora specialists, C. Feuillet and J. M. MacDougal ; no modern cladistic analyses of the family or genus had been published at the time. Cladistic analyses were performed using PAUP* Version 4.0b10 for Macintosh (Sinauer, Sunderland, Massachusetts, USA) with all but one character equally weighted. In the morphological analysis, the absence of petals was given a weight of two merely to enforce the monophyly of the ingroup taxa, as is strongly supported by my DNA-based analysis. Th e morphological and molecular data sets were analyzed using the heuristic search option (MULTREES, SPR, 1000 random replicates, holding fi ve trees per replicate, using the delayed transformation optimization).
Trees were evaluated on the basis of tree length, consistency index (CI), and retention index (RI) as calculated by PAUP*. Bootstrap consensus trees were generated for all data sets (1000 replicates). Congruence of the separate data sets was assessed by comparison of the tree statistics and topologies of the strict consensus trees.

Species concepts
Th e phylogenetic species concept sensu Wheeler and Platnick was primarily employed in this study (Wheeler and Platnick 2000). However, other species concepts such as the biological species concept (Mayr 1942), phenetic species concept (Sokal and Crovello 1970) and autapomorphic concept (Donoghue 1985;Mishler 1985) were also considered and frequently proved useful. In the phenetic analyses of the P. suberosa and P. coriacea complexes, I looked for gaps in the pattern of variation and used sets of morphological characters in species delimitation. I also considered the inability to interbreed (through greenhouse studies conducted by J. M. MacDougal, unpublished data), along with other evidence (easily observed diagnostic morphological characters), as an indication that P. itzensis (J.M. MacDougal) K. Port-Utl. is a distinct entity recognizable at the rank of species. In addition, nearly all of the species in supersection Cieca are cladospecies and possess molecular and morphological autapomorphies (20)(21).

Habit
As noted by MacDougal (1994), the usual size or habit of passionfl owers is seldom recorded by collectors and is poorly known for most species. Th is is especially true of the species in Passifl ora subgenus Decaloba supersection Cieca; the plants are often small in stature and possess small fl owers. According to my own fi eld observations and notes from herbarium specimens the species in the supersection rarely reach a length greater than 8 m. Th ey are perennial climbing or procumbent vines commonly found growing along forest edges.

Stems
Th e pressing and drying which occurs during the making of herbarium specimens causes the stems of most exemplars to appear sulcate, however, observations of living and alcohol-preserved material show that the stems of the species in supersection Cieca are mostly terete; some species have stems that are slightly compressed. In subgenus Decaloba, the stem tip can be cernuous or more or less straight. Th e posture of the stem tip is likely an important taxonomic characteristic in the subgenus and is thought to be under selection by butterfl ies searching for ovipositioning sites (MacDougal 1994). In supersection Cieca, the apices are straight. Th e stems of all species in the supersection are antrorsely appressed-puberulent throughout, with small, unicellular, curved trichomes. Some species are also sparsely to densely pubescent with longer unicellular, rarely multicellular, curved trichomes.

Stipules
In supersection Cieca the stipules are setaceous or narrowly triangular to foliaceous, but foliaceous stipules are found in only four species. Th e stipules that are setaceous or narrowly triangular have only one vein, but those with foliaceous stipules (particularly P. juliana and P. eglandulosa) possess 3-9. Th e stipule margins are always entire.

Leaves
Th e laminas in supersection Cieca, as in the entire genus, are incredibly variable in shape. Th is is likely due to selection pressures from passionfl ower butterfl ies that visually search for particular leaf shapes when looking for ovipositioning sites. Th e leaves may be unlobed or 2-, 3-, or 5-lobed and often exhibit heterophylly (especially heterophyllous species are P. obtusifolia, P. pallida, P. suberosa, and P. tenuiloba). Several species are not heterophyllous and possess bi-lobed (e.g., P. itzensis, P. tacanensis K. Port.-Utl., and P. xiikzodz) or tri-lobed (e.g., P. juliana, P. lancifolia, P. macfadyenii, P. trinifolia, and P. viridifl ora) mature leaves. P. tenuiloba, the Texas longhorn, is the only species in the supersection that possesses leaves that have three or more leaf lobes; the primary lobes of the leaves may also have 2-4 smaller lobes. Th e venation pattern of the leaves in the supersection is usually palmate; even the unlobed leaves of P. pallida possess palmate venation. Th e leaves may also be peltate (especially peltate species are P. coriacea, P. juliana, P. sexocellata, and P. viridifl ora) or not. Th e fi rst few leaves on the poorly known seedlings are usually peltate in most species. Th e margins of the leaves, however, are uniformly entire.

Extrafloral nectaries
Th e leaves of supersection Cieca are simple and commonly bear functional nectaries on the petioles, though two species (P. eglandulosa and P. mcvaughiana) usually do not possess glands. Petioles are typically terete to slightly fl attened. When glands are present on the petioles there are typically two and they are opposite, subopposite or alternate to one another. Th e glands are usually disc-or cup-shaped. Many species in supersection Cieca are characterized by disc-shaped petiolar nectaries that possess edges which are fused to the petiole (P. clypeophylla, P. coriacea, P. itzensis, P. trinifolia, P. viridifl ora, and P. xiikzodz). Others possess cup-shaped nectaries that have raised edges that are not fused to the petiole (P. lancifolia, P. macfadyenii, P. pallida, and P. suberosa subsp. suberosa). Six species and one subspecies have individuals that possess either disc-or cup-shaped nectaries, but one type is more common than another. P. suberosa subsp. litoralis is a widespread subspecies that more commonly possesses cup-shaped nectaries, but there are examples of this species (especially from the South America) that possess disc-shaped nectaries. In P. juliana, P. megacoriacea, P. mcvaughiana, P. obtusifolia, P. sexocellata, and P. tenuiloba, disc-shaped nectaries are much more frequent. Th e positioning of the nectaries can also vary. Most species have nectaries that are found only on the distal half of the petiole (P. itzensis, P. lancifolia, P. macfadyenii, P. mcvaughiana, P. megacoriacea, P. obtusifolia, P. pallida, P. tenuiloba, P. tridactylites, and P. xiikzodz). Others have nectaries that are on the proximal half of the petiole (P. juliana, P. sexocellata, P. tacanensis, and P. viridifl ora). P. suberosa and P. trinifolia are the only two species that can possess nectaries in a variety of positions on the petiole.
Functional laminar nectaries are also present in many species of the supersection. In those species that possess them they occur as submarginal glands associated with minor veins of the abaxial surface. Th e glands are discoid and slightly raised. Th e absence of laminar nectaries is characteristic of P. eglandulosa, P. lancifolia, P. macfadyenii, P. mcvaughiana, P. pallida, and P. tacanensis.

Inflorescence
Shawn Krosnick (2005) has produced the most modern interpretation of the structure of the infl orescence and the evolution of its various forms in the Passifl oraceae; her interpretation follows Troll (1964) andCussett (1968). Basically, the infl orescence is an axillary compound cyme. Various parts of this infl orescence, however, have been reduced. For example, in supersection Cieca and many other species in the family, the peduncle is completely reduced, giving rise to a sessile infl orescence. Th e fi rst order axis of the sessile cyme terminates in a tendril and the second-order side branches terminate in fl owers. Th e prophylls of the fi rst order axis are displaced onto the branches that they normally subtend (Cussett 1968). In most other species in the family, one of the fi rst order prophylls and the two second-order side branch prophylls are retained on each second-order side branch, giving rise to 3-bracteate pedicels collateral with the tendril in the axil of the leaf. Supersection Cieca is unique in that usually none of the prophylls of the fi rst and second-order side branches are retained on the second-order side branches, giving rise to ebracteate pedicels collateral with the tendril in the axil of the leaf. In some species, however, one prophyll (likely from the fi rst order axis) or two prophylls (likely one from the fi rst order axis and one from the second order side branch) are retained on the pedicels. Th us, species of supersection Cieca usually have no fl oral bracts, or up to two; no species have three, as in most Passifl ora. When present the bracts are setaceous and, in some cases, are quickly deciduous.

Flowers
Th e fl owers in supersection Cieca are apetalous and erect or, rarely, positioned horizontally; very rare occurances of one or two well-positioned petals in the otherwise apetalous fl owers have been observed (e.g., P. suberosa subsp. litoralis and P. itzensis). Most fl owers are greenish yellow in color with purplish to reddish markings; two species (P. lancifolia and P. macfadyenii) possess red fl owers. Th e fl owers are small, rarely exceeding 3 cm in diameter; most species possess fl owers that are less than 2 cm wide. Most are bowl-or saucer-shaped, but three species (P. lancifolia, P. macfadyenii and P. viridifl ora) are tubular.

Hypanthium
Th e hypanthium is the portion of the fl ower that holds the nectary and associated structures (the operculum and limen) at its base and bears the perianth, corona, and androgynophore. Th e hypanthium in the fl owers of supersection Cieca is patelliform or dishlike and is less than 3 mm deep, with most species possessing a hypanthium that is less than 1 mm in depth. Th e diameter of the hypanthium is commonly 5-8 mm.

Sepals
Th e species in supersection Cieca possess fi ve, ovate triangular sepals. In most taxa, the sepals are greenish yellow on their outer surfaces, though in P. lancifolia and P. macfadyenii they are red. Adaxially the sepals are greenish yellow (e.g., P. suberosa subsp. litoralis, P. juliana, P. mcvaughiana), red (P. lancifolia and P. macfadyenii), or rarely whitish (e.g, P. suberosa subsp. suberosa). Th e sepals are distinct, except in the tubular fl owers of P. macfadyenii and P. viridifl ora where they are partially connate. In most species, the sepals are refl exed at anthesis. Th e sepals are, on average, 8 mm in length in saucer-shaped fl owers and 15 mm in length in tubular fl owers.

Corona
In supersection Cieca, the corona is mostly in 2 series, a shorter inner series and a (often) much longer outer series. In two species, P. itzensis and P. xiikzodz, the corona is in 7 series; the outer two rows are the longest and the other inner rows are much shorter and nearly equal in form. Occasionally, individuals in the P. suberosa complex (sensu latu) lack an inner coronal row.
Th e fi laments in the outer coronal row are terete, sometimes very slightly capitate, and have a base color of greenish yellow, or purple to red (sometimes very dark reddish purple). Greenish yellow fi laments often have purplish to reddish spots and streaks and may be tipped with bright yellow or white. Red fi laments are often uniform in color or possess yellowish tips. Th e orientation of the outer coronal row is commonly bowlshaped at anthesis, but may be more or less erect or refl exed and fl at.
Th e inner coronal fi lments are usually less than half the length of the outer coronal fi laments and are capitate. Th e inner coronal fi laments are commonly greenish yellow, purplish, or red and have lighter-colored tips. As in the outer coronal row, the greenish yellow fi laments may possess reddish or purplish spots and/or streaks. Th e orientation of the inner fi laments is frequently erect.

Operculum
Th e operculum is considered to be the innermost coronal row in the genus Passifl ora. Th e function of the operculum, however, is generally not to attract pollinators but to cover and protect the fl oral nectary. In supersection Cieca, the operculum is membranous and plicate or, in the nectarless P. itzensis and P. xiikzodz, denticulate. Th e operculum is curved over the nectary and commonly touches the tip of the limen or, particularly in tubular fl owers, completely covers both the nectary and limen and leans against the androgynophore.

Nectary
Th e nectary is positioned at the base of the hypanthium and is a trough that is covered by the operculum. In many species of Passifl ora, there is a raised ring (sensu Jorgensen et al., 1974) or annulus (sensu Tillett 1988) in the nectary trough. A nectar ring or annulus is lacking or very inconspicuous in the species of supersection Cieca; however, in P. tenuiloba the nectary is sulcate. Th e development and physiology of the fl oral necatary of P. eglandulosa [as P. trinifolia] was examined by Durkee et al. (1981). She found that the fl oral nectary development and nectar secretion in this species was similar to that in two other species of Passifl ora (P. warmingii and P. bifl ora) that she studied. Th e activity of an intercalary meristem in the nectary and increased starch deposition in the amyloplasts of the secretory cells parallels the maturation of the nectary phloem and serves as the main source of nectar sugars at anthesis. Th ough she did not measure the sugar concentration of P. eglandulosa, she found that in the other two species in her study the dominant sugar constituent of the nectar was sucrose, with fructose present only in moderate amounts; nectar sugar concentrations are presented below in the discussion on reproductive biology (Durkee et al. 1981).

Limen
Th e limen is a structure that is situated between the nectary and the androgynophore. It is widely considered to be of staminodal origin and, along with the operculum, helps to protect the nectary (Killip 1938;Puri 1948;deWilde 1974). Th e limen in supersection Cieca is adnate to the hypanthium with only its outer edge free. Th e edge is commonly erect and inclined toward the nectary, though in some species it is curved toward the androgynophore. It commonly has a base color of greenish yellow, white, purple or red. When greenish yellow or white, it often possesses reddish spots and/or streaks.

Androgynophore
Th e androgynophore is a central column in the fl ower which consists of an elongate gynophore surrounded by and fused to the staminal fi laments. It is straight in all species of supersection Cieca. It is greenish yellow and often possesses reddish or purplish spots or streaks. It is generally less than 5 mm in height, though tubular species possess androgynophores that reach heights of 25 mm.

Androecium
Th e androecium in supersection Cieca is very uniform. Th ere are fi ve greenish yellow fi laments with versatile and dorsifi xed anthers. In most species the anthers are introrse in bud but fl ip over and are extrorse at dehiscence. In these species, the long axis of the anthers remains parallel to the long axis of the fi laments or, rarely, the long axis of the anthers are perpendicular (or nearly so) to the long axis of the fi laments. Rarely, the anthers only move slightly from the original introrse position, remain introrse, and dehisce distally (upwards). Th e pollen is commonly yellow in color; however, in P. suberosa subsp. suberosa the pollen is light yellow or whitish in color.

Gynoecium
Th e ovary of three carpels is commonly ellipsoid to globose in shape; few species possessing fusiform (P. macfadyenii and P. tridactylites) ovaries. Th e ovary has one locule and the placentation is parietal. Th e ovary possesses a small stipe that extends no more than a millimeter above the adnation of the staminal fi laments. It is more or less glabrous and greenish yellow in color.
Th e styles are slender and free to the base and may be straight or curved. Th ey have a base color of greenish yellow but may possess reddish or purplish spots and/or streaks. Th e stigmas are depressed ovoid and greenish yellow to whitish in color.

Fruits
Th e fruits in the supersection are small (commonly less than 2 cm long) berries that contain one (rarely) to many (80) seeds. Mature fruits are purple or very dark purple with a very thin pericarp. Often the epidermis has a glaucous bloom.

Seeds
Th e seeds in supersection Cieca are compressed and often beaked at the chalazal apex. Th e sculpturing of the seeds is reticulate-foveate. Most species possess 20-30 seeds per fruit. Th e species that have the fewest (<10) seeds per fruit are P. eglandulosa, P. mcvaughiana, and some species of P. pallida and P. suberosa subsp. litoralis. Th e species that commonly possess more than 40 seeds per fruit are P. coriacea, P. juliana, P. sexocellata, and P. viridifl ora. Each pale brown to dark brown seed is surrounded by a fl eshy aril that is somewhat translucent; the aril usually covers only ¾ of the seed. Th e arils that I have tasted are either very mildly sweet or sour.
Passifl ora pallida and Passifl ora suberosa are the only known polyploids in supersection Cieca. Snow and MacDougal (1993) found that P. pallida from Jamaica was a polyploid (tetraploid) with a chromosome number of 2n = 24. Beal (1971) determined that P. suberosa subsp. litoralis from both coastal Argentina and New Guinea had a chromosome number of 2n = 24. He also counted the chromosomes (2n = 24) of a plant of the subspecies from the "U.S.A.", but the locality seems questionable based upon the morphology of the voucher. In addition, Passifl ora suberosa subsp. litoralis does not occur in the wild in the United States, but it is commonly cultivated there. He also found the same chromosome numbers for three clones of P. suberosa subsp. litoralis collected in Australia (Beal 1969(Beal , 1971. Diers (1961) found the diploid chromosome number of 2n = 12 in P. suberosa subsp. litoralis from Lomas de Lachay, Perú. However, I have not been able to locate his voucher specimens (Diers 1961). Storey (1950) also counted the chromosomes of Hawaiian material, which he called P. suberosa. However, I was unable to locate his vouchers and because P. suberosa subsp. suberosa and P. suberosa subsp. litoralis both occur in the Hawaiian Islands, I cannot be certain which subspecies he sampled. However, he did fi nd chromosome numbers of 2n = 24 and 36 in wild populations of the species. He determined that the form with 36 chromosomes was likely an autotriploid derivative of the 24 numbered form. He did not describe the plants that he sampled, but he noted that there were no conspicuous morphological diff erences between the two chromosomal races. He only found that the triploid race had slightly larger leaves and more anthocyanin pigmentation in the young stems and abaxial surfaces of the sepals (Storey 1950).
Th e fruits of supersection Cieca are unilocular berries with thin pericarps that are very dark purple, sometimes with a glaucous bloom. Th ey may contain one (rarely) to many arillate seeds, with the arils mostly clear to slightly opaque and covering one half to three quarters of the seed. Th e fruits also persist on the pedicels for some time after maturity. Van der Werff (van der Werff 1951 and1420) reported that fi nches eat the fruits of P. tridactylites and P. suberosa subsp. litoralis in the Galapagos Islands. Cliff ord Smith (Univ. of Hawaii) has found that the seeds of P. suberosa are dispersed by alien frugivorous birds in Hawaii (http://www.botany.hawaii.edu/ faculty/cw_smith/pas_sub.htm). Th e Mariana fruit bat, Pteropus mariannus mariannus Desmarest, is known to feed on the fruits of Passifl ora pallida on Guam. Passifl ora pallida is a weedy vine there and will grow up into and cover the canopies of forest trees species, especially in disturbed habitats. Feeding by the Mariana fruit bat occurs mostly when the vines grow up in the tops of trees and the bat lands in the tree to feed (Dustin Janeke, pers. comm.; http://www.passionfl ow.co.uk/bats11.htm).

Herbivory
Species of Passifl ora are of particular interest to entomologists, as these plants are larval hosts for passion fl ower butterfl ies (Subfamily Heliconiinae, Family Nymphalidae).
Larvae of the subfamily are almost uniquely restricted to food plants in the Passifl oraceae, giving rise to the name "passion fl ower butterfl ies." Th e close association of species in the Heliconiinae and Passifl oraceae is commonly held up as an example of plant-insect coevolution.
Most of the species of Passifl ora supersection Cieca are utilized by common and widespread species of the subfamily Heliconiinae . Th e known butterfl y herbivores of species of supersection Cieca are listed in Table 1. Most species in the supersection have only one or two known herbivores, but, as one would expect, the species that are widely distributed have a greater diversity of herbivores. Th e more derived species of supersection Cieca (P. xiikzodz, P. juliana, P. viridifl ora, P. coriacea, P. sexocellata, P. megacoriacea) are mainly used by species of Heliconius. Th e early branching species in Table 1. Herbivore records of Heliconiinae for the species of Passifl ora supersection Cieca; "-" = taxa not included in the molecular analysis.  Hawkeswood 1991Benson et al. 1975Benson et al. 1975Benson et al. 1975Minno and Minno 1999P.Schappert, pers. comm. Benson et al. 1975

Agraulis vanillae Dryas iulia Heliconius charitonia
Texas, USA Texas, USA Texas, USA Benson et al. 1975M. Quinn, pers. comm. L. Gilbert, pers. comm. P. viridifl ora Heliconius charitonia Mexico MacDougal 1983 Heliconius erato Belize Meerman 2001 the supersection (P. eglandulosa, P. lancifolia, P. pallida, and P. suberosa subsp. litoralis) are also commonly utilized by Heliconius spp. but are also hosts for early branching genera of the Heliconiinae (Acraea, Agraulis, Dione, Dryas, Dryandula and Philaethria). Passifl ora tenuiloba, which is sister to P. pallida in the morphological analysis but is more closely related to P. coriacea and P. sexocellata in the molecular analysis, serves as a host for Agraulis and Dryas and the more derived genus Heliconius.

Distribution and habitats
Species of supersection Cieca are found from Florida and southern Texas in the United States of America, through Mexico and Central America, from Colombia and Venezuela to Argentina and southern Brazil, and in the Caribbean; they are absent from the Guyana Shield region. Passifl ora pallida and P. suberosa are also found in many areas of the Old World tropics and on many north and south Pacifi c islands to the east of the International Date Line, as the result of introduction by humans. However, the center of diversity is in southern Mexico and northern Central America. Of the 19 species recognized here, fi ve species (P. juliana, P. viridifl ora, P. mcvaughiana, P. tacanensis and P. itzensis) are endemic to Mexico, two to Guatemala (P. clypeophylla and P. trinifolia), two to Jamaica (P. lancifolia and P. macfadyenii), and one to the Galapagos Islands, Ecuador (P. tridactylites). Passifl ora juliana and P. viridifl ora are both found along the Pacifi c coast and in the Pacifi c coastal plain of southwestern Mexico in disturbed tropical deciduous or semideciduous forests of low to moderate elevation (Fig. 43). Th ese two species are not sympatric, with P. juliana found farther north, from areas around the Bahia Chamela in Jalisco to those just south of Manzanillo in southern Colima, and P. viridifl ora occurring from regions just north of Lazaro Cardenas in Michoacan to areas around the Gulf of Tehuantepec in southern Oaxaca. Passifl ora itzensis is found in tropical semideciduous forests from areas near Chichen Itza in Yucatán to localities in southern Quintana Roo north of Chetumal (Fig. 55). Passifl ora mcvaughiana is also found in southwestern Mexico, in high elevation oak, pine/oak or pine forests or montane mesophytic forests on moist hillsides and in barrancas (Fig. 46). Passifl ora tacanensis is known only from three collections in a high altitude tropical montane forest on Volcán Tacaná in Chiapas, Mexico along the border with Guatemala (Fig. 46). Passifl ora clypeophylla has not been found since the type was collected in 1889 (Fig. 39). Based upon locality information included on the herbarium specimens and information gathered by J. M. MacDougal (pers. comm.) on a recent trip to the type locality, P. clypeophylla is a plant of moderate elevation (ca. 1115 m. alt.) and is (or was) likely found on slopes of premontane tropical moist forest. Passifl ora trinifolia is a rare plant found on cliff s and rocks in open, strongly seasonally dry pine and oak forests in northeastern Baja Verapaz,Guatemala (Fig. 39). Passifl ora macfadyenii was last collected in 1979 and repeated attempts to fi nd the plant by myself, Elma Kay (Missouri Botanical Garden), and George Proctor (Institute of Jamaica) have failed. It has been found in tropical dry forests in roadside thickets and wooded limestone hills in the parishes of St. Andrew and St. Th omas (Fig. 32). Passifl ora lancifolia, another Jamaican endemic of supersection Cieca, is found in tropical lower montane mist forests on steep wooded hillsides in the Blue Mountains (Fig. 32). Passifl ora tridactylites is an endemic of the Galapagos and grows in dry tropical forests at altitudes ranging from sea level to 800 m (Fig. 30).
Many of the remaining species of the supersection have wider geographic ranges in Mexico, Central America, and South America. Passifl ora tenuiloba is a plant occurring in arid and semiarid thorn scrub and grasslands from southern Texas to northern Mexico (Fig. 35). Passifl ora xiikzodz is found in the same habitats as its sister P. itzensis, but in addition to being found in the Yucatán Peninsula of Mexico, its range extends to Belize and Guatemala (Fig. 55). Passifl ora sexocellata is found from southern Mexico to Nicaragua (Fig. 52). Th roughout its range this species is found in low, moist to wet tropical forests near streams and rivers, but, in the state of Veracruz, Mexico, it can be found growing on seaside cliff s. Passifl ora megacoriacea is found in Costa Rica and Panama. In the northwestern corner of Costa Rica, in the province of Guanacaste, this species is deeply trilobed and occurs in the premontane transitional belt between the dry tropical forests typical of the Cordillera de Guanacaste and wetter mid-elevation forests (Fig. 50). Th roughout the remainder of its range it is found at lower elevations in dry to wet tropical forests inland and near the sea along the Atlantic and Pacifi c coasts. Passifl ora coriacea is found from northern Colombia and northwestern Venezuela to northern Bolivia (Fig. 49). It occurs in moist to wet tropical forests commonly at elevations of 50-1500 m, reaching higher elevations in the northern part of its range. Passifl ora obtusifolia is found from sea level to 1500 m elevation in tropical deciduous and semideciduous forests in the Pacifi c lowlands and foothills of southwestern Mexico, El Salvador, and Costa Rica; it is found at higher elevations of 500-1500 m in the southern part of its range (Fig. 39). Passifl ora eglandulosa occurs in shady ravines and at the edges of premontane to montane (1500-2800 m) broad leaved forests on volcanic cones from Guatemala to El Salvador and central Honduras (Fig. 37).
Th e species in supersection Cieca with the widest ranges are P. pallida and P. suberosa. Passifl ora pallida has a circum-Caribbean distribution and is found in and along the edges of low elevation, dry tropical forests both inland and near the seashore (Fig. 24). Th is species has also been introduced into the areas of the Old World such as Australia, the Northern Mariana Islands, Comoros, Micronesia, India, Madagascar, Maldives, Mauritius, Palau, the Seychelles, Singapore, the Solomon Islands, and Sri Lanka. Passifl ora suberosa subsp. suberosa is primarily restricted to the islands of the Greater and Lesser Antilles and is found in and along the edges of semideciduous to deciduous, dry to moist tropical forests, both inland and near the seashore, from sea level to 1600 m, but it has also been collected in the Hawaiian Islands, where it is introduced (Fig. 26). In the Greater Antilles, P. suberosa subsp. suberosa is commonly found in and along the edges of moist forests, primarily at higher elevations. It is relatively common on all of the islands of the Greater Antilles, except for Jamaica where it is very rare. In the Bahamas and the Lesser Antilles, it does occur at high elevations but primarily occurs at lower elevations and is found in dry to moist forests. Passifl ora suberosa subsp. litoralis has the widest geographic range of the taxa in supersection Cieca (Fig. 28). It grows in and along the edges of semideciduous to deciduous, dry to moist tropical forests and in secondary successional areas, both inland and near the seashore, from sea level to 2800 m, from northern Mexico, through Central America, to central Argentina and Brazil. In the Old World tropics it has been introduced in Australia, Fiji, New Caledonia, India, Indonesia, South Africa, Spain, Sri Lanka, Taiwan, and Uganda.

Molecular analyses of supersection Cieca
Th e cladistic analysis (PAUP* 4.0b10) of the molecular ITS-1 and ITS-2 data resulted in the generation of three equally parsimonious trees  of 590 steps, a consistency index (CI) of 0.636, and a retention index (RI) of 0.837. Th e topologies presented in these trees are all quite similar, with only minor rearrangements occurring within the P. pallida and P. suberosa clades. Th e strict consensus tree is presented in Fig. 12.
Th e monophyly of supersection Cieca is strongly supported (100% bootstrap). Th ere is also evidence for the monophyly of P. pallida (95% bootstrap), P. lancifolia (87% bootstrap), P. tenuiloba (100% bootstrap), P. sexocellata (93% bootstrap), P. viridifl ora (96% bootstrap), P. juliana (100% bootstrap), P. obtusifolia (50% bootstrap), and P. mcvaughiana (84% bootstrap). Th e Passifl ora coriacea/P. sexocellata and P. xiikzodz J. M. MacDougal/P. itzensis (J. M. MacDougal) K. Porter-Utley clades are monophyletic and well supported with bootstrap values of 100% and 90%, respectively. Passifl ora juliana and P. viridifl ora form a clade (77% bootstrap). A clade consisting of several populations of P. suberosa subsp. suberosa from the Greater Antilles and St. Croix is supported with a bootstrap value of 83%, though this subspecies is, as assessed in this analysis, paraphyletic. A moderately supported clade (73% bootstrap) indicates that an entity of P. suberosa subsp. suberosa from the Caribbean is more closely related to the red, hummingbird-pollinated, Jamaican endemic P. lancifolia than it is to other morphologically similar entities of P. suberosa subsp. suberosa (e.g., P. suberosa subsp. suberosa, St. Croix). . Th e fi rst of three most parsimonious trees from the ITS-1 and ITS-2 data set of Passifl ora supersection Cieca and outgroups. Numbers above branches are branch lengths. Bootstrap values are given below corresponding branches. Tree length = 590; CI = 0.636; RI = 0.837; RC = 0.532. Names followed by the letters "a", "b", "c", "d", and "e" denote clones of the same individual from a particular locality. Figure 10. Th e second of three most parsimonious trees from the ITS-1 and ITS-2 data set of Passifl ora supersection Cieca and outgroups. Numbers above branches are branch lengths. Bootstrap values are given below corresponding branches. Tree length = 590; CI = 0.636; RI = 0.837; RC = 0.532. Names followed by the letters "a", "b", "c", "d", and "e" denote clones of the same individual from a particular locality. Figure 11. Th e third of three most parsimonious trees from the ITS-1 and ITS-2 data set of Passifl ora supersection Cieca and outgroups. Numbers above branches are branch lengths. Bootstrap values are given below corresponding branches. Tree length = 590; CI = 0.636; RI = 0.837; RC = 0.532. Names followed by the letters "a", "b", "c", "d", and "e" denote clones of the same individual from a particular locality. Figure 12. Strict consensus of three most parsimonious trees from the ITS-1 and ITS-2 data set of Passifl ora supersection Cieca and outgroups. Tree length = 590; CI = 0.636; RI = 0.837; RC = 0.532. Names followed by the letters "a", "b", "c", "d", and "e" denote clones of the same individual from a particular locality.
Th e strict consensus tree shows that P. itzensis, P. xiikzodz, P. mcvaughiana, P. juliana, P. viridifl ora, and P. obtusifolia form a clade, with P. obtusifolia sister to the other above-listed species. Within this group, P. juliana, P. viridifl ora, and P. mcvaughiana, three species from southwestern Mexico, constitute a clade. Passifl ora coriacea, P. sexocellata, and P. tenuiloba are also grouped together in all trees. Passifl ora suberosa is nonmonophyletic, but a large number of the accessions of this species do constitute a clade in the strict consensus tree. In addition, P. suberosa subsp. suberosa and P. suberosa subsp. litoralis are also both non-monophyletic. In addition, the cladograms indicate that P. pallida may be sister to the remaining species of supersection Cieca.
Of particular interest are the clones of the various entities of P. suberosa and P. pallida (Fig. 13). Sequences cloned from a single individual of Passifl ora suberosa subsp. suberosa from Haiti are found in two diff erent clades, with two clones ("a" and "b") falling within a moderately supported clade containing other members of the subspecies from the Caribbean and the other two ("c" and "d") forming a group in the strict Figure 13. A portion of the strict consensus tree from the ITS-1 and ITS-2 data set of Passifl ora supersection Cieca and outgroups highlighting cloned entities. Numbers above branches are branch lengths. Bootstrap values are given below corresponding branches. Tree length = 590; CI = 0.636; RI = 0.837; RC = 0.532. Names followed by letters ("a", "b", "c", "d", and "e") and symbols (black square, gray square, white square, and white triangle) denote clones of the same individual from a particular locality. consensus tree that is positioned sister to most of the taxa in the supersection. In addition, cloned entities of Passifl ora suberosa subsp. litoralis from the states of Puebla and Veracruz, Mexico are found in separate clades. Th e clones of the "sub. w/ pall. aff . USA(FL)" entity from Florida occur in both the well-supported P. pallida clade ("a" and "c") and the clade containing P. suberosa along with the rest of the species from the supersection ("b" and "d"), indicating that there is gene fl ow, likely resulting from hybridization, between P. suberosa and P. pallida.

Phenetic analyses of species complexes
Th e Passifl ora suberosa Complex. A principal components analysis (PCA) of the 51 character morphological data set ( Table 2) for the P. suberosa complex is presented in Fig. 14. Taxa recognized in this revision as P. suberosa subsp. suberosa, P. suberosa subsp. litoralis, P. tridactylites, and P. pallida were included in the analyses and are labeled accordingly. In addition, entities that may be of hybrid derivation and possess both P. pallida and P. suberosa affi nities are indicated. Principal components I, II, and III account for 48.3%, 14.2%, and 7.7% of the variation, respectively, for a total of 70.2%. Principal component axis I is most highly infl uenced by (presented in decreasing order of component loadings) ( Table 3): (1) length of the lateral leaf lobe, (2) distance from the outline of the leaf to the margin of the leaf sinus, and (3) leaf width; axis II by (1) androgynophore length, (2) sepal length, and (3) stipe length; axis III by (1) petiole length, (2) distance of the petiolar nectaries from the petiole base, and (3) number of laminar nectaries. Th e PCA plots of axes I and II and axes I and III separate P. tridactylites from the other taxa in the analysis, but the remaining taxa are poorly separated. Th e fi rst principal component (PC1) consists primarily of information from vegetative characters and is largely an indication of leaf size and leaf lobe depth. Th e second principal component (PC2) has low component loadings for the vegetative characters and high component loadings for the fl oral characters and is primarily an indicator of fl ower size. Because PC1 and PC2 are by defi nition not correlated, this division of fl oral and vegetative characters between these fi rst two PCs indicates that there is little correlation between fl oral and vegetative characters among the entities in the P. suberosa complex.
A PCA analysis of the data set for the P. suberosa complex based on 31 fl oral characters (Table 2) is presented in Fig. 15. Principal components I, II, and III account for 47.1%, 15.0%, and 9.5% of the variation, respectively, for a total of 71.6%. Principal component axis I is most highly infl uenced by (presented in decreasing order of component loadings) ( Table 4): (1) androgynophore length, (2) sepal length, and (3) length of the fi laments in the outer coronal row; axis II by (1) number of fi laments in the inner coronal row, (2) pedicel length, and (3) number of fi laments in the outer coronal row; axis III by (1) pedicel length, (2) stipe length, and (3) nectary width. Th e PCA plots of axes I and II and axes I and III clearly separate P. tridactylites, P. suberosa subsp. suberosa, P. suberosa subsp. litoralis, and P. pallida. Th e fi rst and third principal Table 2. Characters used in the morphology-based principal components analysis of the Passifl ora suberosa complex (* = characters used in the PCA analysis of the data set for P. suberosa complex based on fl oral characters) All measurements were recorded in mm. For a discussion of character state delimitation see Methods.
*Number of fi laments in outer coronal row 10.
*Length of fi laments in outer coronal row 11.
*Width of fi laments in outer coronal row 12.
*Number of fi laments in inner coronal row 13.
*Length of fi laments in inner coronal row 14.
* Distance from leaf outline to sinus margin 43.
Depth of leaf lobe 44.
Leaf width components are a measure of overall fl ower size and PC2 is mostly a measure of the relationship among the number of fi laments present in each coronal row. When PC1 and PC3 of fl oral characters are plotted for individual plants (Fig. 15), individuals from the same species and subspecies tend to cluster together although there is limited N Measurements 45.
Length of pubescence on leaf 46.
Length of pubescence on stem 49.
Tendril width Figure 14. Principal components analysis of the data set for the P. suberosa complex based on 51 morphological characters ( Table 2).  Figure 15. Principal components analysis of the data set for the P. suberosa complex based on 31 fl oral characters ( Table 2).

74.
Leaves with primary veins diverging and branching at base (0); leaves with primary veins diverging and branching above base (1); leaves with secondary veins forming a series of loops (2) 75. Laminar nectaries absent (0); laminar nectaries present (1) 76. Stem entirely greenish yellow (0); stem with reddish purple coloration (1) 77. Stipules glabrous (0); stipules pubescent (1) Table 6. Character values used in the construction of the neighbor joining tree for taxa in the    (Tables 5-6). is 5.9 mm; char. #19) and longer outer coronal fi laments (the average length of the fi laments in the outer coronal row of P. tridactylites is 7.5 mm and the average length of the fi laments in the outer coronal row of this accession of P. suberosa subsp. litoralis from the Galapagos is 5.2 mm; char. #8). Passifl ora suberosa subsp. suberosa forms two distinct clusters, a Lesser Antillean group and a largely Greater Antillean group, with one accession from Hawaii (Oahu) and one from St. Croix found within the Greater Antillean cluster.
Most of the P. pallida accessions form a cluster toward the "base" of the neighbor joining tree, but four accessions of P. pallida are placed elsewhere, intermixed with P. suberosa subsp. litoralis, indicating that the diff erences between these two taxa are sometimes diffi cult to discern. Th ere are many smaller clusters of P. suberosa subsp. litoralis, all primarily positioned adjacently on the tree between the P. tridactylites/P. suberosa subsp. suberosa group and the P. pallida group. Th ese results suggest that both P. tridactylites and P. suberosa subsp. suberosa may have evolved from a nonmonophyletic P. suberosa subsp. litoralis. One small cluster of two accessions from coastal areas of Ecuador and Peru are located in a small group just "below" the larger P. tridactylites/P. suberosa subsp. suberosa cluster. However, the rest of the clusters of P. suberosa subsp. litoralis contain individuals from mainly high elevations in Mexico and Central America with South American accessions scattered within them. In addition, two individuals, which may be of hybrid origin and possess affi nities of P. pallida and P. suberosa, occur at two diff erent locations in the neighbor joining tree, but they are both most similar to P. pallida.
Th e Passifl ora coriacea Complex. A principal components analysis (PCA) of the entire morphological data set (Table 7) for the P. coriacea complex is presented in Fig. 17. Species that are recognized in this revision as P. coriacea, P. megacoriacea, and P. sexocellata are included and labeled accordingly. Principal components I, II, and III account for 30.8%, 19.2%, and 10.5% of the variation, respectively, for a total of 60.5%. Principal component axis I is most highly infl uenced by (presented in decreasing order of component loadings) ( Table 8): (1) degree that the leaf is peltate, (2) stipe length, and (3) androgynophore length. Axis II is most highly infl uenced by (1) degree that the leaf is peltate, (2) leaf width, and (3) length of the lateral leaf lobe, and axis III by (1) pedicel length, (2) number of fi laments in the inner coronal row, and (3) length of the pubescence on the stem. Th e PCA plots of axes I and II and I and III separate P. megacoriacea from P. coriacea and P. sexocellata. Th e fi rst principal component (PC1) and the third principal component (PC3) consist of information from both fl oral and vegetative characters. Th e second principal component (PC2) has low component loadings for the fl oral characters and high component loadings for the vegetative characters and is primarily an indicator of the degree that the leaf is peltate and leaf size. Th e graphs in Fig. 17 place individuals with larger fl owers that have short fl oral stipes and long pedicels and larger leaves that are less peltate with narrower lateral lobes in the right half of the scatter plots.
A PCA analysis of the data set for the P. coriacea complex based on fl oral characters (Table 7) is presented in Fig. 18. Principal components I, II, and III account for Table 7. Characters used in the morphology-based principal components analysis of the Passifl ora coriacea complex (* = characters used in the PCA analysis of the data set for the P. coriacea complex based on fl oral characters) All measurements were recorded in mm. For a discussion of character state delimitation see Methods.
1. *Pedicel length 2. *Pedicel width 3. *Stipe length 4. *Stipe width 5. *Hypanthium diameter 6. *Sepal length 7. *Sepal width 8. *Number of fi laments in outer coronal row 9. *Length of fi laments in outer coronal row 10. *Width of fi laments in outer coronal row 11. *Number of fi laments in inner coronal row 12. *Length of fi laments in inner coronal row 13. *Width of fi laments in inner coronal row 14 16.6%, and 12.0% of the variation, respectively, for a total of 70.1%. Principal component axis I is most highly infl uenced by (presented in decreasing order of component loadings)( Table 8): (1) stipe length, (2) androgynophore length, and (3) sepal length; axis II by (1) pedicel length, (2) number of fi laments in the inner coronal row, and (3) androgynophore width; axis III by (1) stipe length, (2) number of fi laments in the inner coronal row, and (3) nectary width. Th e PCA plots of axes I and II and I and III indicate that all three species, P. megacoriacea, P. coriacea, and P. sexocellata, are phenetically separable. Individuals of the same species tend to cluster together although there is only a small amount of overlap among entities of P. coriacea and P. sexocellata. Th e pattern refl ected in both plots in Fig. 19 strongly relates to the overall Figure 17. Principal components analysis of the data set for the P. coriacea complex based on 44 morphological characters (Table 7). size of the fl ower, with entites in the right half of the scatter plot having larger fl owers than those individuals in the left half. Th e unrooted neighbor joining tree produced from an analysis of the entire morphological data set ( Table 9-10) of the P. coriacea complex is shown in Fig. 19. Th e accessions representing P. megacoriacea and P. coriacea, the two most morphologically distinct taxa in the P. coriacea complex, are each clearly clustered toward the "top" of the neighbor joining tree. Accessions representing P. sexocellata are clustered toward the "base" of the tree, but two accessions representing P. sexocellata (from Belize and Mexico) are more similar to P. megacoriacea than to other members of P. sexocellata from Mexico and Central America. However, P. megacoriacea can be easily separated  (Table 7). Table 9. Component loadings for axes I, II, and III from a principal components analysis of the P. coriacea complex (Fig. 18) Th e values were computed from quantitative fl oral variables. from these accessions by its elongated androgynophore (the average androgynophore length of P. megacoriacea is 7.5 mm and the average androgynophore lengths of the accessions of P. sexocellata from Belize and Mexico are between 3 and 5 mm, respectively; char. #15).
Outer coronal fi laments dark reddish purple at base with yellow apex (0); outer coronal fi laments purplish at base and whitish toward tips (1) 35. Limen erect or inclined toward the operculum (0); limen recurved (1) 36. Nectary without raised annulus (0); nectary with raised annulus (1) 37. Leaves transversely elliptic (0); leaves distinctly trilobed and ovate in general outline (1) 38. Leaves not variegated (0); leaves variegated (1) analysis, branch support is low for most of the tree. Only six of the branches have bootstrap scores >50%, and only three have values where some confi dence can be obtained: the branch grouping P. arbelaezii L.Uribe and P. lancetillensis J.M.MacDougal Figure 19. Unrooted neighbor joining tree resulting from the analysis of the morphological data from entities within the P. coriacea complex based upon 38 morphological characters (Tables 10-11). Table 11.
Character values for taxa used in the phenetic analysis of the Passifl ora coriacea complex ( Fig. 19) A = 0/1; ? = condition unknown.
Leaf venation with primary veins diverging and branching at base (0); leaf venation with secondary veins forming a series of loops (1); leaf venation with primary veins diverging and branching above base (2) 84.
Plants & J.Meerman has a bootstrap of 92%, the branch grouping P. lobbii Mast. subsp. lobbii and P. exoperculata Mast. has a bootstrap of 95%, and the branch grouping P. itzensis and P. xiikzodz has a bootstrap of 100%.
Th e presence/absence of petals (character # 45, state 2) was given a weight of two in this analysis (all other characters having a weight of one) in order to insure that the ingroup is resolved as monophyletic; the monophyly of the supersection has already been confi dently established (100% bootstrap value) with molecular sequence data. Th e monophyly of supersection Cieca is supported by the following apomorphies: the absence of petals, styles that are less than 6 mm in length (#15, 1), outer coronal fi laments that are commonly less than 5 mm in length (#8, 4), fewer than 35 fi laments present in the inner coronal row (#10, 2), staminal fi laments that are frequently less than 5.5 mm long (#13, 1), and 0-2 fl oral bracts (#42, 1). Th e absence of petals is the only nonhomoplasious character supporting the monophyly of the supersection. Within supersection Cieca, Passifl ora clypeophylla, P. juliana, P. viridifl ora, P. megacoriacea, P. mcvaughiana, P. xiikzodz, P. itzensis, P. sexocellata, P. coriacea, P. tacanensis, and P. obtusifolia form a clade. Within this group, P. juliana, P. viridifl ora, and P. megacoriacea constitute a clade, with Passifl ora juliana and P. viridifl ora being most closely related. A clade consisting of P. mcvaughiana, P. xiikzodz, P. itzensis, P. tacanensis, P. sexocellata, and P. coriacea is also evident, with P. xiikzodz and P. itzensis present as sister species. Th e two Guatemalan endemics, P. trinifolia and P. eglandulosa, form a clade. A clade consisting of P. tridactylites, P. lancifolia, and P. macfadyenii is also present, with P. lancifolia and P. macfadyenii sister to each other. Passifl ora tenuiloba, P. pallida, and P. suberosa subsp. litoralis form a clade, with P. tenuiloba and P. pallida being most closely related. Finally, Passifl ora suberosa subsp. suberosa is cladistically basal within the supersection.
Passifl ora suberosa subsp. suberosa is defi ned by the presence of white sepals (#49, 1), fi laments in the inner coronal row that are reddish purple at the base with a yellow capitate head (#56, 2), whitish pollen (#64, 1), a sulcate fl oral nectary fl oor (#71, 1), ovoid fruits (#76, 3) and three-lobed leaves (rarely unlobed or bilobed) (#84, 4). Th e remaining members of the supersection form a clade based upon a shift from plants that commonly possess four or more laminar nectaries to those that possess none (#37, 2)(with shifts to leaves with less than four and more than four laminar nectaries occurring in many taxa) and ellipsoid ovaries (#65, 1)(with several shifts to globose and fusiform ovaries in several taxa).
Within Passifl ora supersection Cieca P. lancifolia, P. macfadyenii, P. tridactylites, P. tenuiloba, P. pallida, and P. suberosa subsp. litoralis form a clade . Th is group possesses petiolar nectaries that are positioned on the upper half of the petiole (#36, 1) and androgynophores that possess red or purple pigmentation (#59, 1). Th e taxa commonly lack laminar nectaries, but when laminar nectaries are present they are submarginal glands that are associated with minor veins of the abaxial surface and never occur proximal to the lateral leaf veins (#89, 6). Passifl ora lancifolia and P. macfadyenii form a clade with 66% bootstrap support. Passifl ora lancifolia and P. macfadyenii are both Jamaican endemics and possess the following synapomorphies: sepals that Table 13. Character values for taxa used in the morphological cladistic analysis of   Table 13.
Th e remaining taxa in this analysis occur in two major clades. With the exception of two taxa (P. megacoriacea and P. obtusifolia) all of the species that have been separated out of the P. coriacea complex (Passifl ora tacanensis, P. mcvaughiana, P. itzensis, P. xiikzodz, P. sexocellata, and P. coriacea s.s.) form one major clade. Th e second major clade consists of P. juliana and P. viridifl ora, along with P. megacoriacea. Th ese two major groups form one larger clade that is delimited by the following synapomorphies: more than 40 fi laments in the outer coronal row (#7, 1), ellipsoid ovaries (#65, 0), seeds that are more than 3.8 mm in length (#26, 1), and the presence of more than four laminar nectaries (#37, 0).

Discussion of all cladistic and phenetic analyses
Th e morphological and molecular analyses presented here confi rm the monophyly of Passifl ora supersection Cieca. In addition, both phenetic and cladistic analyses have increased our understanding of some of the complex biological issues infl uencing the evolution of the group.
Congruence between phylogenetic hypotheses generated from independent data sets, when subjected to reliable methods of phylogenetic analysis, is often thought to be evidence for considering those hypotheses as representative of the "true" phylogeny. Confl ict may indicate theoretical or procedural problems in one or both of the analyses, or that additional data are needed to resolve the phylogenetic relationships in question (Hillis 1987). Th ere is a considerable amount of incongruence between the molecular and morphological phylogenies for Passifl ora supersection Cieca presented in previous chapters. Sample size has likely had an infl uence on this incongruity. For the morphological data set, specimens for all the species of the supersection from throughout their geographic ranges were carefully measured and examined in order to determine the extent of variation of the characters. However, such a large sample size in the molecular study was not feasible due to limited sample availability and expense of analysis. Th ere were also several polymorphic sites evident in the ITS sequences for several of the species within supersection Cieca, further increasing the need for more infraspecifi c sampling. Most importantly, the complex evolutionary history (involving hybridization and polyploidy) of several of the entities within the group has undoubtedly had a signifi cant eff ect on both data sets, increasing the amount of incongruity in the analyses.
In an attempt to overcome the diff erence in sample sizes between the morphological and molecular data sets, additional cladistic analyses were undertaken in which the operational taxonomic units were reduced to those for which both molecular and morphological data were available. Separate analyses of the reduced morphological and molecular data sets resulted in the production of four equally parsimonious trees that were 224 steps long and 1,638 equally parsimonious trees that were 332 steps in length, respectively. Th e consensus trees from both of these analyses were compared and were conspicuously incongruent in their topologies. In addition, there were signifi cant changes in the resulting phylogenies, as compared to the original analyses, that were likely due to decreased (and likely inadequate) taxon sampling. I have chosen not to combine my morphological and molecular data sets because of this incongruence and the knowledge that complex biological processes are likely infl uencing the pattern of diversity within Passifl ora supersection Cieca.
On a more positive note, the phylogenetic hypotheses based on the morphological and molecular data sets for supersection Cieca agree in several respects. In the molecular analysis, there is evidence for the monophyly of P. pallida (95% bootstrap), P. lancifolia (87% bootstrap), P. tenuiloba (100% bootstrap), P. sexocellata (93% bootstrap), P. viridifl ora (96% bootstrap), P. juliana (100% bootstrap), P. obtusifolia (50% bootstrap), and P. mcvaughiana (84% bootstrap) . In the morphological analysis, each of these species is also diagnosable by unique combinations of character states . Th e molecular and morphological analyses also agree in their support for the monophyly of a clade containing P. itzensis and P. xiikzodz and a clade composed of P. juliana and P. viridifl ora.
Th e sister-group relationship between P. xiikzodz and P. itzensis is highly supported in both the molecular (90%) and morphological (100%) trees. In the morphological analysis P. xiikzodz and P. itzensis are diagnosable by a number of morphological characters. In addition, crossing studies by MacDougal (1992) suggest that these species are unable to interbreed, whereas fruits with viable seeds were easily produced between two clones of P. xiikzodz (MacDougal 4690 and MacDougal 4677) (C. Feuillet, pers. comm.). However, though the two species are separated from one another in the strict consensus tree in the molecular analysis, there is no bootstrap support for either of the species-level clades. Despite the lack of statistical support for P. itzensis and P. xiikzodz as cladospecies in the molecular analysis, I have elevated P. xiikzodz subsp. itzensis and P. xiikzodz subsp. xiikzodz to the rank of species based on their consistent diff erences in fl oral morphology. Th e way in which P. itzensis displays its pollen is dramatically diff erent from that of P. xiikzodz, indicating a shift in pollinators. Th us, P. xiikzodz and P. itzensis are considered sibling species, which do not appear to be able to interbreed and possess consistent and easily observed diagnostic morphological characters.
Both the molecular and morphological data support the monophyly of a clade containing P. juliana and P. viridifl ora; each is considered a cladospecies, and together they compose a moderately supported (77%) clade in the molecular analysis. In his original description of P. juliana, MacDougal (1992) discussed its similarities with P. viridifl ora, and both species are found in similar habitats along the Pacifi c coast and in the Pacifi c coastal plain of southwestern Mexico. Th e primary diff erences between these two species are the shape of the stipules and several changes in fl oral and vegetative mor-phology associated with a shift in pollinators (hymenopteran to hummingbird pollination). Again, each of these two species is well-supported with bootstrap values greater than 95% in the molecular analysis and are morphologically diagnosable cladospecies.
Th e phylogeny based on the molecular data is diff erent from that based on the morphological data in many ways (compare . Passifl ora mcvaughiana is most closely related to P. juliana and P. viridifl ora in the DNA-based trees. In the morphological analysis, P. mcvaughiana is in a clade with P. itzensis, P. xiikzodz, P. sexocellata, P. coriacea, and P. tacanensis because they all possess bilobed or shallowly trilobed, transversely elliptic leaves and very similar fl owers, especially in the case of P. coriacea and P. sexocellata. Passifl ora obtusifolia is sister to the clade containing P. itzensis, P. xiikzodz, P. mcvaughiana, P. juliana, and P. viridifl ora in the molecular cladogram. However, in the morphological analysis, P. obtusifolia is sister to all of the species listed above (P. itzensis, P. xiikzodz, P. mcvaughiana, P. juliana, and P. viridifl ora) plus P. megacoriacea, P. sexocellata, P. coriacea, and P. tacanensis. Passifl ora coriacea, P. sexocellata, and P. tenuiloba form a clade in the molecular analysis. In the morphological analysis, P. tenuiloba is most closely related to P. pallida, a species that also occurs in southwestern Texas and northeastern Mexico. In the molecular analysis, P. coriacea is sister to P. sexocellata and they form a monophyletic group. In the morphological analysis P. coriacea is also sister to P. sexocellata, but they are in a clade with P. mcvaughiana, P. xiikzodz, and P. itzensis. Based upon the molecular data, Passifl ora lancifolia is sister to P. suberosa subsp. suberosa but is more closely related to P. macfadyenii, P. pallida, and P. suberosa subsp. litoralis in the morphological analysis. In the molecular analysis P. pallida is sister to all of the members of the supersection, whereas P. suberosa subsp. suberosa is positioned as sister to the other taxa in the morphological analysis (see . Clearly, additional molecular data from more variable gene regions are needed to help resolve phylogenetic relationships in Passifl ora supersection Cieca. Several gene regions were sequenced in an attempt to attain independent sets of molecular information, but none proved variable enough to resolve the phylogeny (trnL-trnF, cytosolic-expressed glutamine synthetase, G3pdh, psbA-trnH). Th e only region that proved promising was waxy (granule-bound starch synthase), but up to seven copies of the gene are found within diploid individuals from the supersection and my preliminary results (not shown) could not be interpreted without extensive and additional sampling.
Th e Passifl ora coriacea Complex. Jussieu described P. coriacea in 1805 from a specimen collected in Colombia by Bonpland. Shortly afterward several authors (e.g., Smith 1814 and Kunth 1817) described additional species based upon characters of the leaves (P. clypeata Sm. and P. diff ormis Kunth). However, their descriptions seem to give only an account of the vegetative variation within diff erent populations of P. coriacea Juss. in Colombia. Schlechtendal (1854) described P. sexocellata and diff erentiated it from both P. coriacea and P. diff ormis based primarily upon the vegetative morphology of the species (because the fl owers of P. coriacea were largely unknown), but his careful description of the fl owers of his new species diff ered markedly from those in Kunth's brief description of P. diff ormis (for further information please see discussion of P. sexocellata below). In his revision of the American species of Passifl oraceae, Killip (1938) listed all of the species discussed above plus several others (P. obtusifolia and P. cheiroptera Cortés) in synonomy under P. coriacea. However, since that revision, MacDougal (1992MacDougal ( , 2001 described two new species, P. xiikzodz and P. mcvaughiana, and resurrected one previously described species, P. obtusifolia, that were reported from southwestern Mexico under the name P. coriacea. In addition, both the phenetic and cladistic analyses based upon both molecular and morphological data presented here support the recognition of three distinct taxa: P. coriacea, P. sexocellata, and P. megacoriacea. Th e multivariate statistical analysis of the quantitative morphological characters for P. coriacea s. l. produced a plotting pattern that clearly supports the delimitation of the previously undescribed species P. megacoriacea, but it is not until an analysis of the fl oral characters alone was undertaken that P. coriacea and P. sexocellata became phenetically separable. Neighbor joining analyses of both the qualitative and quantitative morphological characters also support the recognition of these three taxa, with only two accessions representing P. sexocellata (from Belize and Mexico) appearing more similar to P. megacoriacea than to other members of P. sexocellata from Mexico and Central America. Th e morphological cladistic analysis of the supersection suggests that P. sexocellata and P. coriacea evolved from a common ancestor, but that P. megacoriacea is more closely related to P. juliana and P. viridifl ora. Passifl ora megacoriacea is placed with these species primarily based upon characters relating to an increase in fl ower size (ovary size, operculum length, etc.) that might be attributed to independent shifts to larger pollinators. However, vegetative and reproductive characters other than those relating to overall fl ower size (e.g., the shape of the fl owers with an erect outer corona that is bent toward the androgynophore, leaf shape, etc.) in P. megacoriacea suggest a closer relationship with P. coriacea and P. sexocellata. Unfortunately, I was unable to obtain material of P. megacoriacea for DNA sequencing, and a molecular analysis that includes P. megacoriacea is needed and will likely settle the issue. Nevertheless, it is clear in the molecular and morphological cladistic analyses that P. coriacea and P. sexocellata are sister to each other (with a bootstrap value of 100% in the DNA-based tree) and are both clearly diagnosable cladospecies, and it is also likely that P. megacoriacea is closely related to these two taxa.
Th e Passifl ora suberosa Complex. Th e oldest herbarium specimens that I have seen of any of the members of the P. suberosa complex were collected in the 1700s, and since that time plant collectors have deposited thousands of specimens of P. suberosa s. l. (3,244 of which I annotated) in herbaria around the world. Linnaeus (1745, 1753) originally described four species that have, over the years, been considered to be part of (at specifi c and/or subspecifi c levels) the P. suberosa complex. Since that time, various systematists engaged in revising the genus and family have described new species (e.g., Cavanilles 1790; Hooker 1867), varieties and subvarieties (e.g., Roemer 1846; Masters 1872) or lumped various entities under one species name (Killip 1938). In his revision of the American species of Passifl oraceae, Killip (1938) recognized P. suberosa in the widest sense and considered the various species, subspecies, and varieties falling within this species to be too intergrading and indistinct to merit taxonomic recognition. He concluded that P. suberosa was an extremely variable species and that no constant characters permitted the maintenance of the proposed variants as distinct taxa. Since Killip's revision, the variability in gross morphological characters, as seen in herbarium specimens of P. suberosa s. l., the over-reliance of many authors on the vegetative morphology in sorting out entities within the "species," and the sheer task involved in sorting through the thousands of specimens collected from around the world, has helped to perpetuate his broad concept of this species. However, both the phenetic and cladistic analyses based upon both molecular and morphological data presented here support the non-monophyly of P. suberosa s. l. and the recognition of four distinct taxa within this complex: P. suberosa subsp. suberosa, P. suberosa subsp. litoralis, P. pallida, and P. tridactylites.
In the multivariate analyses of the quantitative morphological characters for P. suberosa s. l., little correlation was found between fl oral and vegetative characters. Instead, the variability in many of the vegetative characters in the complex made it diffi cult to elucidate distinct taxa. However, an analysis of the fl oral characters alone produced an ordination pattern that supports the delimitation of P. pallida, P. suberosa subsp. suberosa, P. suberosa subsp. litoralis, and P. tridactylites, though some overlap among entities of P. suberosa subsp. litoralis and P. pallida does exist. Neighbor joining analyses of both qualitative and quantitative morphological characters also support the recognition of four taxa, but accessions of P. pallida were intermixed with accessions of P. suberosa subsp. litoralis, further indicating that the diff erences between these two taxa are sometimes diffi cult to discern. Th e results of the neighbor joining analysis also suggest that both P. tridactylites and P. suberosa subsp. suberosa may have evolved from P. suberosa subsp. litoralis. Th e morphological cladistic analysis of the supersection as a whole indicates the converse, that is, P. suberosa subsp. suberosa is sister to the rest of the species in the supersection. Th e position of P. suberosa subsp. suberosa in the cladistic analysis is questionable. Nevertheless, it is placed there because it shares a number of characters (e.g., sepal color) with the chosen outgroups which are probably actually derived within the supersection. Passifl ora suberosa subsp. litoralis, P. pallida, and P. tridactylites are in a clade with P. tenuiloba, P. lancifolia, and P. macfadyenii in the morphological analysis. Passifl ora pallida and P. suberosa subsp. litoralis are present in a clade with P. tenuiloba, underscoring the close relationship between P. pallida and P. suberosa subsp. suberosa, with P. tenuiloba forming a clade with P. pallida based upon the width of the fl oral nectary. Passifl ora tridactylites is placed in a clade with P. lancifolia and P. macfadyenii, which is likely a consequence of the increased fl ower size and other adaptations to non-hymenopteran pollinators in these three species. Th e molecular analysis indicates that P. lancifolia is sister to P. suberosa subsp. suberosa, though with only moderate support (73%), and according to my morphological analysis, P. macfadyenii is sister to P. lancifolia. In addition, P. tridactylites is very similar in many morphological characters to P. suberosa subsp. litoralis, both of which occur on islands of the Galapagos, underscoring their probable close relationship. Passifl ora tridactylites, P. lancifolia, and P. macfadyenii are similar in many aspects to P. suberosa, and all three likely evolved from it. However, it is doubtful that they are each other's closest relatives.
In the molecular cladistic analysis, some of the ambiguities apparent in the results from the phenetic analyses and morphology-based cladistic analyses were resolved or at least clarifi ed. In all three trees resulting from an analysis of the ITS sequence data of supersection Cieca, P. pallida appeared to be monophyletic with bootstrap support of 95%, but P. suberosa subsp. suberosa and P. suberosa subsp. litoralis are not indicated as monophyletic. Th eir non-monophyly is likely due, at least in part, to some amount of gene exchange between these entities. While the amplifi cation of the ITS region yielded a single product for all accessions of the supersection as revealed by gel electrophoresis, I noticed that the directly sequenced PCR product for several accessions of P. suberosa subsp. suberosa, P. suberosa subsp. litoralis and what morphologically appeared to be P. pallida contained polymorphic sites (where two discernible peaks of approximately equal strength appeared in the chromatograms). Th us, I began to clone several of my PCR products and found that the accessions with polymorphic sites often proved to possess diff ering, apparently functional copies of ITS; there were no signifi cant nucleotide substitutions, insertion-deletion events, or substitutions (particularly in conserved regions) apparent in the sequences that would indicate that the copies were nonfunctional. Four individuals of supersection Cieca contained polymorphic sites (two accessions of P. suberosa subsp. litoralis from Puebla and Veracruz, Mexico, one accession of P. suberosa subsp. suberosa from Haiti, and one accession from Florida, USA that had small fl owers similar to P. pallida), while the rest of the species did not contain any polymorphisms in the ITS region. Th e cloned sequences of Passifl ora suberosa subsp. suberosa from Haiti were placed in two diff erent clades, with two clones falling within a moderately supported clade containing other members of the subspecies from the Caribbean and the other two forming a group in the strict consensus tree, which is positioned sister to P. suberosa subsp. litoralis. In addition, cloned entities of Passifl ora suberosa subsp. litoralis from the states of Puebla and Veracruz, Mexico are found in separate clades. Th e clones of the accession that morphologically fi ts the description of P. pallida ["sub. w/ pall. aff . USA (FL)", see Fig. 13] from the United States occur in both the well-supported P. pallida clade and the clade containing P. suberosa along with the rest of the species from the supersection, indicating that there is gene fl ow, likely resulting from hybridization, between P. suberosa and P. pallida. Th is gene fl ow has likely obscured the distinctiveness of P. pallida and contributed to the broad circumscription of P. suberosa.
As shown in the phenetic analyses of P. suberosa s. l., there is clearly some overlap in the morphological characters of P. pallida and P. suberosa. Th is may indicate that there is limited gene fl ow occurring between these species, and the molecular data are consistent with this hypothesis. However, most specimens of P. suberosa and P. pallida are clearly separable and the inclusion of P. pallida, a well-supported cladospecies, within the circumscription of P. suberosa would render this species extremely nonmonophyletic and obscure the distinctiveness of an early divergent lineage within the supersection. According to the molecular data there defi nitely seems to be gene fl ow between the subspecies of P. suberosa, and though they are morphologically distinct, I felt it best to treat these two somewhat geographically isolated taxa at the subspecifi c level (as opposed to the species level). Th e exact impact that hybridization and polyploidy are having on the evolution of P. pallida and P. suberosa remains unknown, but it is clear that these processes have blurred the distinctions between these species and made the sorting out of phylogenetic relationships within these widespread and variable taxa extremely diffi cult. However, population level studies incorporating cytological data and DNA fi ngerprinting likely would reveal their consequences, clarifying the circumscription of P. suberosa and its subspecies.
Phenetic and cladistic analyses of the supersection based on morphological and molecular characters were utilized to generate hypotheses of species phylogenetic relationships and redefi ne specifi c entities, especially within the two species complexes. Th e phylogenetic analyses presented here confi rm the monophyly of the supersection. In the molecular and morphological analyses each of the species of the supersection, with the exception of P. suberosa, is monophyletic and diagnosible by a unique combination of character states. Th ere is support in the molecular and morphological analyses for the monophyly of a clade containing P. itzensis and P. xiikzodz and a clade composed of P. juliana and P. viridifl ora. In addition, there is strong support in the molecular analysis for a clade comprising P. coriacea and P. sexocellata.
Four taxa that were formerly included in the P. suberosa complex are recognized here: P. pallida, P. suberosa subsp. suberosa, P. suberosa subsp. litoralis, and P. tridactylites. Both the molecular and morphological analyses show that Passifl ora suberosa is not monophyletic, a situation that may be quite common in plants, suggesting that a criterion of monophyly for species recognition may be inappropriate. It is quite possible that P. suberosa has been caught in the paraphyletic "stage" of speciation, and the data indicate that it might be more logical to view the phylogenetic status of a species as a property that may change over time. Peripherial isolate speciation, such as that which presumably has given rise to the Galapagos endemic, P. tridactylites. Passifl ora tridactylites possesses a fl ower that appears to be adapted to a larger pollinator than its relative P. suberosa, which is mainly pollinated by hymenopterans, and P. tridactylites may be moth pollinated. Th e analyses also indicate that there is limited gene fl ow, likely in the form of hybridization, occurring between P. suberosa and P. pallida. Th is gene fl ow has obscured the distinctiveness of P. pallida, a species that is likely sister to the remaining members of the supersection, and has contributed to the traditional broad circumscription of P. suberosa. Passifl ora pallida and P. suberosa have also been shown to be polyploids (including triploid, tetraploid and hexaploid counts). Th e exact impact that hybridization and polyploidy are having on the evolution of P. pallida and P. suberosa is unknown; however, it is clear that these processes have blurred the distinctiveness of these two species and made the sorting out of phylogenetic relationships between and within them very diffi cult.
Th ree species from the P. coriacea complex are recognized: P. coriacea, P. megacoriacea, and P. sexocellata. It is clear in the molecular analysis that P. coriacea and P. sexocellata are sister to each other, and both are clearly diagnosable. It is also likely that P. megacoriacea is closely related to these two taxa. Passifl ora megacoriacea and P. sexocellata were recognized as specifi cally distinct as a result of this investigation.
Passifl ora xiikzodz and P. itzensis are recognized at the specifi c level, as opposed to the subspecifi c level, due to consistent diff erences in fl oral morphology which likely resulted from a shift in pollinators. Based upon fl oral morphology, the majority of the species in the supersection are probably pollinated by insects, likely hymenoptera. However, pollination by hummingbirds has also been reported for several species of the supersection and appears to have evolved at least twice, once in P. viridifl ora and again in the common ancestor of P. lancifolia and P. macfadyenii, leading to dramatic shifts in fl oral form in these species. Th e species of supersection, for the most part, are not sympatric and where two or more species coincide, they are found growing at different elevations or in diff erent habitats. Several species within the supersection (e.g., P. lancifolia, P. macfadyenii, P. tridactylites, and P. viridifl ora) fi t the peripheral isolate model of speciation and have developed divergent ecological amplitudes that have allowed them to invade novel habitats and exploit diff erent spectrums of pollinators.
Most of the species of Passifl ora supersection Cieca are utilized by common and widespread species of the subfamily Heliconiinae. Many of the species in the supersection have only one or two known herbivores, but, as one would expect, the species that are widely distributed have a greater diversity of herbivores. Th e extent and nature of mutual descent between the species of supersection Cieca and the Heliconiinae, still remains largely unknown.
Lastly, the utilization of only one concept to defi ne the species of supersection Cieca was inadequate. However, meaningful biological entities were identifi ed through the integration of elements from several concepts (e.g., the biological, phenetic, autapomorphic and diagnostic species concepts), along with information from many new taxonomic collections, observations of living material, and detailed phenetic and phylogenetic analyses (based on DNA and/or morphological data).
Description. Herbaceous or woody, perennial (rarely annual or with annual shoots from perennial roots), tendril-climbing vines or lianas, rarely shrubs or small trees lacking tendrils; usually containing cyanogenic glycosides having a cyclopentenoid ring system; glabrous to densely pubescent with simple trichomes, rarely glandheaded. Stems terete to lobed or sharply angled, occasionally with anomalous secondary growth, the shoot apex erect to cernuous. Leaves alternate (very rarely subopposite to opposite), simple (rarely palmately compound), petiolate, often with variously shaped and positioned extrafl oral nectary glands on the petiole; laminas unlobed or lobed, often heteroblastic, pinnately to often palmately (rarely pedately) veined, variegated or not, entire to serrate, peltate or not, often bearing small nectaries associated with marginal teeth or indentations, or abaxially submarginal, or abaxial between the major veins. Stipules setaceous or narrowly triangular to foliaceous, persistent or early deciduous, entire to serrate, sometimes the margins with glands, occasionally cleft. Tendrils axillary, simple (rarely compound), representing a modifi ed fl ower stalk of the central part of the infl orescence, straight, curved, or circinate during development at the shoot apex, rarely with adhesive terminal disks. Infl orescences axillary, bracteate or rarely ebracteate, cymose, the central pedicel developed into a tendril, the peduncle very reduced or usually absent, the pedicels then arising collateral to the tendril (sometimes aborted), solitary or paired; secondary infl orescences may be present as condensed axillary or terminal shoots, determinate or rarely indeterminate; pedicels articulate distal to bracts, the distal portion called the fl oral stipe; bracts setaceous and scattered to foliaceous or pinnatifi d and involucrate, occasionally glandular at margin. Flowers bisexual (sometimes functionally staminate), actinomorphic or rarely the reproductive parts zygomorphic; hypanthium ± fl at to campanulate, occasionally the perianth basally connate/adnate into a fl oral tube; sepals 5 (very rarely 8), quincuncially imbricated (rarely non-overlapping) in the bud, occasionally carinate, sometimes with a subapical projection; petals 5 (very rarely 8) or sometimes wanting, quincuncially imbricated (rarely nonoverlapping) in the bud, the same length as or shorter (rarely slightly longer) than the sepals; corona present at the base of the calyx or corolla or adnate to the inside of the fl oral tube, in 1 to many series of distinct to occasionally connate, short to elongate, often showy fi laments or outgrowths, sometimes membranous, the innermost series, called the operculum, often connate at least basally, frequently membranous and shielding the nectary; the limen (extrastaminal nectariferous disk) present as a ring or cup around base of androgynophore (or rarely the ovary if androgynophore absent), or discoid or conical and adnate to the fl oor of hypanthium. Stamens 5(8 in one species), usually alternate with the petals, borne on an often elongate androgynophore or androgynophore rarely absent; fi laments free just below ovary or rarely connate into a tube around ovary; anthers introrse in bud, moving to become extrorse (rarely latrorse) at anthesis, dorsifi xed, versatile, dehiscing longitudinally, borne parallel or perpendicular to their fi laments; pollen binucleate, 3-to 12-colporate. Carpels 3(-5), connate, ovary superior, unilocular, borne on an often elongate androgynophore (rarely sessile), placentation parietal, anatropous ovules numerous on each placenta; styles distinct, rarely connate near base; stigmas capitate, clavate, reniform, or occasionally bilobed. Fruit a few to many seeded berry, rarely a loculicidal or anomalously dehiscent capsule. Seeds arillate, usually fl attened, the testa pitted, reticulate-foveate, or transversely grooved or sulcate; endosperm slightly ruminate, oily, abundant; embryo straight, the cotyledons usually elliptic to oblongelliptic; germination epigeal (rarely hypogeal). Chromosome numbers: n = 6, 9, 10, 12 (rarely 7, 11, 18, 42 Description. Small to medium-sized climbing or procumbent vines with perennial stems from woody perennial rootstocks or taproots, antrorsely appressed-puberulent more or less throughout, with unicellular, curved or occasionally erect trichomes, and sometimes sparsely to densely pubescent with longer unicellular, rarely multicellular, curved trichomes. Stems terete to somewhat compressed and two-edged, the shoot apex erect. Leaves simple, commonly bearing nectaries on the petiole (except in P. eglandulosa and P. mcvaughiana); petioles sometimes canaliculate, biglandular (rarely eglandular or with only a single gland) with opposite, subopposite or alternate, discoid, cupulate, obconical or capitate extrafl oral nectaries; laminas unlobed or 2-to 3-lobed (rarely 5-lobed), often exhibiting heterophylly, sometimes cordate at base, entire (very rarely crenate), venation palmate, variegated or not, peltate or not, sometimes bearing small abaxial disciform or crateriform nectaries present ± submarginally between the major veins (very rarely associated with leaf crenations). Stipules setaceous to foliaceous, persistent, narrowly to widely ovate, rarely oblong or obovate, symmetrical or sometimes asymmetrical, entire, not glandular. Tendrils simple, lacking adhesive disks, straight or slightly curved during development at shoot apex. Infl orescences sessile in leaf axils, the pedicels solitary or paired, collateral with tendril, articulate, the articulation generally several mm below the fl ower; secondary infl orescences sometimes present as condensed axillary or usually terminal shoots, determinate or usually indeterminate; bracts 1-2 or lacking, narrowly ovate to entire. Flowers erect or rarely ± horizontal, greenish yellow sometimes with purplish to reddish markings, or red, hypanthium usually shallow, occasionally the calyx basally connate into a conspicuous fl oral tube; sepals ovate-triangular, not corniculate, greenish yellow, red, or rarely whitish; coronal fi laments in 2 series (rarely 1 or 7 series), greenish yellow, sometimes with yellow and/or purple to red markings, or purple to red (sometimes very dark reddish purple), linear, often subcylindrical in cross-section, inner fi laments usually capitate; operculum connate, membranous, plicate (very rarely denticulate), incurved or rarely semierect and laying against androgynophore; nectary trough-shaped or rarely absent, commonly lacking or possessing a very inconspicuous nectar ring or annulus; limen adnate to fl oor of hypanthium or rarely absent (in P. viridifl ora the limen present as a shallow cup around base of androgynophore), the edge commonly erect and inclined toward the nectary, rarely curved toward the androgynophore. Staminal fi laments with the free portions actinomorphic; anthers commonly extrorse at anthesis with their axes maintained parallel, rarely perpendicular, to the fi lament or rarely the anthers move only slightly from the original introrse position, remain introrse, and dehisce distally (upwards); pollen ellipsoid to spherical, 6-syncolporate. Carpels 3; ovary ellipsoid or globose, rarely slightly ovoid, obovoid or fusiform, glabrous or rarely densely pubescent with curved, unicellular or rarely multicellular trichomes; styles slender, less than 1.5 mm in diameter; stigmas capitate, depressed-ovoid. Fruit a one (rarely) to manyseeded purple or very dark purple berry, arils pale-translucent covering approximately 3/4 of the seed. Seeds more or less compressed, often beaked at chalazal apex, reticulate-foveate. Germination epigeal. Chromosome numbers: n = 6 (12,18). Commonly lacking c-glycosylfl avones and usually containing fl avonol 3-O-glycosides. Fig. 22 Key to the species of Passifl ora supersection Cieca Description. Slender, climbing, perennial vine 1-7 m long or more, sparsely to densely pubescent with unicellular curved trichomes on petiole, leaf, stem, and stipule, 0.20-0.30(-0.7) mm long, 0.02-0.03 mm wide, also minutely antrorsely appressedpuberulent throughout with unicellular, curved trichomes, 0.06-0.11 mm long, 0.02-0.03 mm wide. Flowering stems 0.6-1.6(-2.5) mm in diameter, terete or somewhat compressed, greenish yellow to very dark reddish purple, with the base woody and cork-covered. Stipules 2.1-6.9 mm long, 0.2-0.9 mm wide, narrowly ovate-triangular, sometimes slightly falcate, acute; petioles 0.3-1.8(-2.9) cm long, with 2 (rarely 1), opposite to alternate, stipitate or sometimes sessile, slightly obconical to capitate nectaries (very rarely crateriform), 0.3-0.8 mm wide (on the widest axis), 0.2-1.1 mm high, borne in the distal half of the petiole (0.49-0.92 of the distance from the base toward the apex of the petiole). Laminas 1. 8-8.8(-12.  dish purple or reddish purple or very dark reddish purple, limen fl oor 1.6-2.6 mm in diameter, greenish yellow or greenish yellow fl ushed with reddish purple or reddish purple or very dark reddish purple; androgynophore (1.7-)2.2-3.5 mm long, 0.4-0.9 mm wide, greenish yellow or greenish yellow with a fl ush of reddish purple at base or greenish yellow with reddish purple spots and streaks or very dark reddish purple; free portions of the staminal fi laments 1.4-3.0 mm long, 0.2-0.4 mm wide, linear, greenish yellow; anthers 1.1-1.9 mm long, 0.5-1.3 mm wide, pollen yellow; styles 1.6-4.3 mm long including stigmas, 0.1-0.4 mm wide, greenish yellow; stigmas 0.5-1.2 mm in diameter; ovary 1.1-1.8 mm long, (0.7-)1.0-1.5(-1.9) mm wide, ellipsoid to globose, greenish yellow. Berry 7.6-9.5 mm long, 6.9-8.8 mm in diamater, globose, or ellipsoid, very dark purple (5P 2.5/2). Seeds (4-)8-24(-33), 2.8-3.5 mm long, 1.9-2.2 mm wide, 1.1-1.4 mm thick, obovate in outline, acute at both ends, reticulate-foveate with each face marked with ca. 12-20 foveae; germination type epigeal.
Phenology. Flowering and fruiting throughout the year. Distribution. In the New World tropics: Central America, Mexico, United States (Florida and Texas), Venezuela, and the West Indies. Introduced in the Old World tropics: Africa, Asia, and Australia. Growing in shrubs, trees or trailing on the ground in secondary successional areas and along the edges of dry tropical forests, both inland and near the seashore, primarily at low elevations but sometimes occurring at elevations as high as 800 m. Commonly associated with calcareous/alkaline substrate.
Ethnobotany. In Réunion, the fruits may be used as a substitute for ink (Jean Jacques, pers. comm.).
Discussion. Passifl ora pallida as recognized here exhibits a substantial amount of morphological variation across its range. Th e various forms that the leaves may take have led to the proposal of many species and varietal names. For example, a plant of this species may possess only unlobed leaves, only trilobed leaves, or leaves that are unlobed, bilobed, and trilobed. Th is type of variation can be seen throughout the range of this species. However, the fl owers of P. pallida are diagnostically small, with a narrow hypanthium, short sepals, short coronal fi laments, and narrow fl oral nectaries.
Th e only species with which P. pallida may be confused is P. suberosa. Passifl ora pallida is vegetatively similar to both P. suberosa subsp. litoralis and P. suberosa subsp. suberosa, and without fl owering material these taxa can be diffi cult to distinguish. Th e position of the petiolar nectaries has often been used to separate species in closely related taxa in Passifl ora. However, though the petiolar nectaries are generally located closer to the petiole apex in P. pallida than in the South American populations of P. suberosa subsp. litoralis and P. suberosa subsp. suberosa, the upland Mexican/Central American populations of P. suberosa subsp. litoralis also have petiolar nectaries positioned very near the petiole apex. Th e leaf base of P. pallida is commonly not cordate. Passifl ora suberosa possesses leaves that are frequently cordate, though this character is somewhat variable in the upland Mexican/Central American populations of P. suberosa subsp. litoralis. Th ough foliage color is diffi cult to discern from herbarium specimens, my experience in the fi eld and photos taken of P. pallida and P. suberosa in the fi eld by others show that P. pallida commonly possesses leaves that are paler in color and often less lustrous than P. suberosa. Reproductive structures are more reliable in separating P. pallida and P. suberosa. Th e hypanthium of P. pallida is commonly 2.8-4.1 mm in diameter and the inner coronal fi laments are usually less than 1.5 mm long. In P. suberosa s. l., the hypanthium is commonly 4.0-8.8 mm in diameter and the inner coronal fi laments are frequently 1.5-3.9 mm long. Th e outer coronal fi laments are also short, less than 4.0 mm in P. pallida, and although that overlaps with the 2.5-8.1 mm range observed in P. suberosa, the character is frequently observable in herbarium specimens. Where the distributions of P. pallida and P. suberosa overlap in the Antilles, P. pallida is typically found in and along the edges of subtropical and tropical forests at or near sea level (rarely exceeding 200 m), whereas P. suberosa commonly occurs in and along the edges of tropical forests above 500 m.
Th e most common variant of P. pallida (as exemplifi ed by E. Killip 41876, on Sugarloaf Key, Monroe Co., Florida, USA; E. Cabrera 1475, S of Akumal, Quintana Roo, Mexico; and J. Tillich 3558, in Black River, Mauritius) has ovate leaves that may be unlobed, bilobed or trilobed on the same plant. When unlobed, the leaves are commonly greater than 2.0 cm wide. When lobed, the leaves are usually shallowly lobed 0.20-0.41 the distance to the base, the lateral and central lobes are greater than 1.0 cm wide, and the angle between the lateral lobes is 45-100°. Another less common variant of P. pallida (as exemplifi ed by J. K. Small & C. Mosier 5511, from Cox Hammock, Miami-Dade Co., Florida, USA; and J. Small & J. Carter 194,between Perrine and Long Prairie, Miami-Dade Co., Florida, USA), has narrowly ovate leaves that may be unlobed, bilobed or trilobed on the same plant. When unlobed, the leaves are commonly less than 1.0 cm wide. When lobed, the leaves are usually deeply lobed 0.82-0.90 the distance to the base, the lateral and central lobes are commonly less than 0.7 cm wide, and the angle between the lateral lobes is greater than 100°. However, all the specimens brought together here as P. pallida are all relatively small in stature in their native habitats in the New World, possess similar small fl owers with short coronal fi laments and occur in a similar range of elevations.
MacDougal has reported the appearance of an occasional, well-formed but small petal in other species within supersection Cieca (MacDougal, 1992). I have also seen this in P. pallida in several of my greenhouse accessions and in the fi eld in Quintana Roo, Mexico. Stegmaier (1973) reported that Dasiops passifl oris McAlpine (Diptera: Family Lonchaeidae) infests the fruits of P. pallida in southern Florida. He found that the female fl y oviposits on the fruits and the larvae feed on the arils and fruit fl esh. In this study, in which he collected a total of 1040 wild passion fruits from P. pallida occurring on a single farm in Hialeah, Florida, he also found that the mature fruit may contain from 4 to 17 seeds per fruit (Stegmaier 1973).
Passifl ora pallida is a pest plant where it occurs in many areas of the Old World. In New Guinea, Neville Kemp reports that the probable disperser of P. pallida is the longtailed macaque or crab-eating macaque (Macaca fascicularis) (Kemp and Burnett 2003).
In Linnaeus' 1753 edition of Species Plantarum, he describes three small-fl owered entities, P. pallida L. ("Habitat in Dominica, Brasilia"), P. hirsuta L. ("Habitat in Dominica and Curassao") and P. minima L. ("Habitat in Curassao"), for which the historical references include phrases such as "fl ore minore" (P. pallida L.), "fl ore & fructu minimis" (P. hirsuta) and "fl ore fl avescente omnium minimo" (P. minima L.). Charles Wright (1869), in an article discussing the genus Jussiaea L., chose the name P. pallida over P. minima. In the article he commented on the "embarrassing" status of the species of Passifl ora and the unwise reliance upon vegetative morphology in species circumscription within the Cuban species of Passifl ora (Wright, 1869:480). In the article he states, "I have lately carefully examined the Cuban species called P. minima, hederacea, pallida, angustifolia, suberosa, &c., and come to this conclusion:-P. pallida, L., is an old and appropriate name, to which belong P. minima, L., and P. angustifolia, Sw., certainly; P. hederacea, Cav., P. suberosa L., probably; and, from the description, I judge P. lineariloba, Hook. f. to be only another form of it." It is possible that Passifl ora hirsuta was not considered by Wright in his article because of the confusion surrounding its circumscription (see below) or because he had not encountered the taxon in Cuba.
In the 1753 edition of Species Plantarum, Linnaeus indicated that he was wellacquainted with P. pallida and refers to the diagnosis and drawing in his Dissertatio botanica de Passifl ora (1745), that shows an unlobed, ovate leaf with two petiolar nectaries positioned near the apex of the petiole. Linnaeus cites an illustration by Plumier (pl. 89, in Description des plantes de l'Amérique 1693) that also exemplifi es his P. pallida. However, he also refers to a fi gure by Morison (1680) that shows a plant with a large fl ower that possesses sepals and petals (likely in the subgenus Passifl ora) with unlobed, ovate leaves. In the 1745 dissertation, Hallman specifi cally states that the fl owers of P. pallida L. are "pentapetala", referring to the lack of petals; this decision was based upon the careful comparison of diagnoses from other petalous taxa in the treatment. An examination of the Linnaean herbarium (microfi che) did not reveal an herbarium specimen that could reasonably be attributed to the species described as P. pallida by Linnaeus. Th ere is one specimen in the Linnaean herbarium labeled P. pallida, but it is a post 1753 accession that represents a large-fl owered taxon from subgenus Passifl ora. Th ough there is a small amount of confusion surrounding P. pallida L., largely attributable to Linnaeus' reference to Morison's illustration and the post-Linnaean accession referred to above, it is clear from the diagnoses in the 1753 edition of Species Plantarum and the 1745 dissertation, that Linnaeus was referring to a plant that had unlobed, ovate leaves and small, pale, apetalous fl owers (Jarvis 2007). Th e lectotype of P. pallida L. (designated here) is Plumier's plate in Description des plantes de l'Amérique (1693) in which he illustrated several entities of both P. pallida and P. suberosa subsp. suberosa. Incidentally, Linnaeus chose the epithet, pallida, to refer to the pale-colored fl ower. Th ough the fl owers are frequently pale in color, they may also be highly colored.
Linnaeus (1753) also describes Passifl ora minima L. as a trilobed plant in which the central lobe is longer than the lateral lobes. He cites the diagnosis and drawing in the 1745 dissertation that shows a plant with narrowly trilobed leaves that lack petiolar glands. Linnaeus (1753) also refers to a fi gure by Plukenet (1696) that closely matches Linnaeus' diagnoses and the drawing in the dissertation. Neither Linnaeus nor Hall-man described the fl owers of P. minima L., but it can be inferred by the historical references in the dissertation that the fl owers were small and lacked petals. In Killip's treatment of P. suberosa, he states that there are two sheets of P. minima from the "West Indies" of uncertain origin in the Linnaean Herbarium and designated them "type of P. minima" (1938:93) without specifying one of the sheets specifi cally. According to Jarvis (2007), the lectotype of P. minima L. (designated by Wijnands 1983) is specimen 1070.20 (LINN). Th e lectotype closely matches Linnaeus' diagnoses and the drawing in the dissertation and possesses small fl owers apparently lacking petals (as observed on a microfi che of the herbarium). However, the lectotype of P. minima is a very unusual example of the small-fl owered entity, as the lack of petiolar nectaries in this taxon is very rare.
Passifl ora hirsuta L. has been the source of confusion for several taxonomists of Passifl ora, and under his treatment of Passifl ora foetida var. moritziana (Planch.) Killip ex Pulle, Killip (1938) discussed the problem. Linnaeus (1753) cited several references in his treatment of P. hirsuta, often with accompanying illustrations, that undoubtedly refer to P. foetida. However, he also referred to an illustration by Plumier (pl. 88, in Description des plantes de l'Amérique 1693) that is clearly P. pallida. As in his other species descriptions, he also cites the diagnosis and drawing in the 1745 dissertation by Hallman that shows a trilobed, densely pubescent leaf with rather large petiolar nectaries that are positioned on the distal half of the petiole and, thus, cannot be P. foetida as this species lacks petiolar nectaries. Th e diagnosis in the 1753 edition of Species Plantarum is unclear. However, in the 1745 dissertation Hallman states that the fl owers of this taxon are pale and small, the involucre is lanceolate, and the fruits are deep blue. Hallman goes on to say that the taxon that he is describing is somewhat similar to the next (P. foetida L.) but diff ers in that the fl owers are opposite (paired) and the involucre consists of only a single bract. Hallman is clearly describing one of the entities in the P. suberosa complex, as the fl owers are commonly paired in the leaf axils, members of the species complex do sometimes possess one or two lanceolate bracts, and the fruits are very dark purple. In Passifl ora foetida only one fl ower is present in the leaf axils, the involucre consists of three large bracts that are pinnatifi d or pinnatisect, and the fruits are yellow to red. Th ough the leaf as illustrated in the dissertation is distinctly cordate and broadly ovate, which is a bit unusual for P. pallida, Linnaeus' reference to Plumier's drawing leads me to conclude that P. hirsuta L. is a synonym of P. pallida. It is also the only original material that corresponds to the current concept of the species.
Phenology. Flowering and fruiting throughout the year. Distribution. In the New World tropics. Introduced in the Old World tropics. Growing in shrubs and trees or trailing on the ground in secondary successional areas, along the edges of semideciduous to deciduous, dry to wet tropical forests, both inland and near the seashore, 0-2500 m.

Key to the subspecies of Passifl ora suberosa
Phenology. Flowering and fruiting throughout the year. Distribution. Th roughout the West Indies. Introduced in the Hawaiian Islands. Growing in shrubs, trees or trailing on the ground in secondary successional areas, along the edges of semideciduous to deciduous, dry to moist tropical forests, both inland and near the seashore, 0-1600 m.
Discussion. In the Greater Antilles, P. suberosa subsp. suberosa is commonly found in and along the edges of moist forests, primarily at higher elevations. It is relatively common on all of the islands of the Greater Antilles, except for Jamaica, where it is very rare. In the Lesser Antilles, it does occur at high elevations but primarily occurs at lower elevations and is found in dry to moist forests.
Th e vegetative morphology of P. suberosa subsp. suberosa is incredibly variable. Nevertheless, throughout most of its range the subspecies commonly has trilobed leaves at reproductive nodes; only ca. 10% of the specimens examined have leaves that are unlobed at all nodes. Approximately 20% of the specimens possess unlobed, bilobed and trilobed leaves on sheets of the same collection. Th e leaves of Passifl ora suberosa subsp. suberosa are commonly lobed less than 50% of the distance from the outline of the leaf to the leaf base and the lateral lobes are ½ -¾ the length of the central lobe. Th e leaves are frequently dark green on their adaxial surfaces and have cordate bases. Th e juvenile leaves of P. suberosa subsp. suberosa are often peltate and frequently possess laminar nectaries; however, the leaves on older plants are only very rarely peltate and usually do not have nectaries. Th e vegetative parts of the plant also possess varying amounts of reddish purple pigmentation, and the stems and new growth are often entirely reddish purple. Passifl ora suberosa subsp. suberosa is relatively small in stature, rarely exceeding a length/height of fi ve or six meters in the fi eld. Th e fl owers are more than 2.5 cm in diameter, with white sepals, coronal fi laments that are dark reddish purple with yellow apices and whitish pollen. Th e fruits are usually ovoid and very dark purple.
In the Lesser Antilles, there are three morphological variants. One of these variants occurs in the Grenadines and has large leaves (over 10 cm wide) that are deeply trilobed (more than half the distance from the leaf outline to the leaf base) with long, commonly oblong lateral lobes that are at least three quarters the length of the lateral lobe. Th e leaves are often distinctly peltate and frequently possess four laminar nectaries (two on either side of the central leaf vein and one proximal to each lateral vein). Another variant occurs primarily in Dominica and Martinique and has deeply trilobed leaves with wider, ovate lateral lobes and deeply cordate bases. Th e leaves are not as large as the fi rst variant (ca. 5-8 cm wide), but possess four laminar nectaries in the same positions as the entity in the Grenadines. Th e last variant occurs on several of the Windward Islands and has trilobed leaves with ca. 10 laminar nectaries. Th e nectaries are positioned near the leaf margin, creating crenations where they appear and are commonly positioned proximal to the lateral leaf veins, a very rare condition in the subspecies. All of these Lesser Antillean forms have the longest fl oral stipes and sepals in the subspecies. In the Dominican Republic and Cuba there is an additional variant that has unlobed leaves at all nodes. Th e leaves are exceptionally long for the subspecies (>10 cm), more coriaceous and possess petiolar nectaries that are wider and somewhat discoid, as opposed to the cupulate or capitate condition common in the subspecies.
Passifl ora suberosa subsp. suberosa is sympatric with three species in supersection Cieca: P. pallida, P. lancifolia, and P. macfadyenii. It can be easily separated from P. lancifolia and P. macfadyenii using both vegetative and reproductive characters. Th e most obvious features are that the leaves of P. macfadyenii and P. lancifolia are very densely pubescent with long, unicellular curved trichomes, whereas P. suberosa subsp. suberosa appears glabrous (i.e., primarily microscopically antrorsely appressedpuberulent). Th e fl owers of P. macfadyenii and P. lancifolia are also tubular and possess bright red sepals. Passifl ora suberosa subsp. suberosa has the cup-shaped fl owers typical of the supersection and white sepals. However, P. pallida and P. suberosa subsp. suberosa can be diffi cult to separate without reproductive material. Th e leaves of P. suberosa subsp. suberosa are darker green in color than those of P. pallida and sometimes have laminar nectaries, these strictly absent in P. pallida. Th ey are also wide, i.e., (2.9-)5.0-12.0(-17.1) cm, in P. suberosa subsp. suberosa, and although this overlaps with the (0.3-)6.0-7.0(-10.6) cm range in P. pallida, the character can frequently be used to distinguish between them. In addition, the leaf bases of P. suberosa subsp. suberosa are cordate (when they are not peltate), whereas those of P. pallida are very rarely cordate and usually are acute to cuneate. Th e stems, leaves (especially at their margins), tendrils, and stipules are frequently reddish purple in P. suberosa subsp. suberosa, and the vegetative parts of P. pallida generally possess little, if any, reddish purple coloration. Passifl ora pallida may be densely pubescent where it occurs in the Caribbean, but P. suberosa subsp. suberosa appears glabrous. Th e fl owers of P. pallida are much smaller than those of P. suberosa subsp. suberosa. Passifl ora pallida has sepals that are very rarely greater than 8 mm long, but the sepals of P. suberosa subsp. suberosa are always longer than 8 mm. Th e hypanthium in P. pallida is 2.2-4.2 mm wide, whereas that of P. suberosa subsp. suberosa is 5.5-8.8 mm wide. Passifl ora pallida has short staminal fi laments (1.4-3.0 mm), and P. suberosa subsp. suberosa has staminal fi laments that are 3.4-6.8 mm long. Th e sepals of P. suberosa subsp. suberosa are white, whereas those of P. pallida are commonly greenish yellow; though P. pallida may possess light colored sepals in the Yucatán Peninsula of Mexico. Th e fruits of these taxa are also quite diff erent; P. suberosa subsp. suberosa usually has ovoid fruits and P. pallida has globose or ellipsoid fruits. In the Greater Antilles, P. suberosa subsp. suberosa is commonly found at higher elevations and in more mesic habitats than P. pallida. In other areas in the world their habitats are less distinct, but the species can be distinguished morphologically.
Cliff ord Smith in the Dep. of Botany at the University of Hawaii reports that P. suberosa subsp. suberosa, as recognized here, is a minor weed in Hawaii in subcanopy layers where it smothers shrubs, small trees and the ground layer. In some areas it can also smother the upper canopy layer. He has also found that the seeds are dispersed by alien frugivorous birds.
Discussion. Passifl ora suberosa subsp. litoralis has the widest geographic range of any species in supersection Cieca. In the New World, its range extends from northern Mexico, through Central America, to central Argentina. In these areas it may be confused with P. pallida and P. obtusifolia, which are sometimes similar vegetatively. Th e similarities and diff erences between these two species are discussed under their respective descriptions. Th e primary diff erence between P. pallida and P. suberosa subsp.
litoralis is the hypanthium diameter, with that of P. pallida rarely exceeding a width of 4.0 mm and that of P. suberosa subsp. litoralis commonly 4.0 mm or wider. One of the more useful characters employed in separating P. obtusifolia and P. suberosa subsp. litoralis is the presence/absence of infl orescences. When mature, Passifl ora obtusifolia bears fl owers in long infl orescences (i.e., 5.3-18.3 cm) and P. suberosa subsp. litoralis almost always lacks infl orescences; when P. suberosa subsp. litoralis does possess infl orescences they are not as long (i.e., 2.0-5.0 cm).
Th ere are three major morphological variants of Passifl ora suberosa subsp. litoralis. In Mexico and Central America, P. suberosa subsp. litoralis possesses shallowly trilobed leaves (commonly less than half the distance from the leaf outline to the leaf base) with the length of the central lobe often greatly exceeding that of the lateral lobes and an angle between the lateral veins that is frequently between 40° and 80°. Th e lateral lobes are also oblong to elliptic. Th e broadly capitate petiolar nectaries are commonly positioned on the distal half of the petiole, often over 0.60 the distance from the base to the apex of the petiole. Th e leaf bases are often cuneate to acute but rarely cordate. In Mexico and Central America, Passifl ora suberosa subsp. litoralis is often found in high elevation (1000-3000 m) moist pine and oak forests along streams and rivers, but it may also occur in very dry forests with cacti (e.g., Cephalocereus forests of Tehuacán) and other species common in matorral vegetation (e.g., Tamaulipan matorral).
On the western side of South America (Colombia to Peru and Argentina), P. suberosa subsp. litoralis possesses leaves very much like those of the Mexico/Central American variant, but the petiolar nectaries may be more discoid and are commonly posi- tioned on the proximal half of the petiole. Th e lateral lobes are commonly distinctly ovate and diverge at an angle of 80-100°. Th e leaf bases are also distinctly cordate. In this region it is found in low (near sea level and on cliff s above the sea) to high (to 3000 m) elevation moist forests commonly along streams and rivers, but it also occurs in tropical dry forests.
On the eastern side of South America, in southeastern Brazil, the leaves are commonly trilobed but may also have unlobed, bilobed or trilobed leaves present on the same plant. Th e petiolar nectaries are commonly discoid and positioned on the proximal half of the petiole. Th e lateral lobes are ovate, but longer than those common in western South America, and commonly diverge at an angle of greater than 100°. Th e leaf bases are distinctly cordate. In Brazil, P. suberosa subsp. litoralis is more common in coastal dunes and tropical dry forests, but it does occasionally occur in higher elevation moist forests as well. Th is variant is the only form of P. suberosa subsp. litoralis found in the Old World. Laminar nectaries are commonly present in all three of the these variants.
Phenology. Flowering and fruiting throughout the year. Distribution. Endemic to the Galapagos Islands. Growing in shrubs, trees or trailing on the ground in secondary successional areas and in dry tropical forests with Castela, Scalesia, Psidium, and Bursera, 0-800 m.
Discussion. Passifl ora tridactylites may be confused with P. suberosa subsp. litoralis, which also occurs in the Galápagos Islands. Both species exhibit a great amount of variation in their vegetative morphology, with both species possessing all of the diff erent vegetative forms described by Hooker, and I have not been able to fi nd any vegetative characters that can reliably be used to distinguish between them. However, the fl owers and fruits of these two species are quite diff erent. Th e sepals of P. tridactylites are commonly 10-14 mm long, whereas those of P. suberosa subsp. litoralis do not exceed a length of 10 mm. Th e outer coronal fi laments are long, more than 6.6 mm, in P. tridactylites, and the fi laments in P. suberosa subsp. litoralis are commonly less than 6.0 mm long. Th e androgynophore in P. tridactylites is diagnostically long, more than 8.0 mm, whereas that of P. suberosa subsp. litoralis is always less than 6.0 mm. Passifl ora tridactylites has long fusiform fruits, exceeding 12.8 mm. Th e fruits of P. suberosa subsp. litoralis are 7.1-11.9 mm long and ellipsoid to globose. According to Lawesson (1988), the habitats of these two species are diff erent, with P. tridactylites occurring in dry lowland areas and P. suberosa subsp. litoralis in mesic habitats. John MacDougal (pers. comm.) found abundant Lepidopteran scales on the inside of several fl owers of pressed P. tridactylites specimens, indicating visits by butterfl ies and/or moths and thus a probable shift in pollinators as a likely selective force leading to the clear fl oral diff erences in these two species. Van der Werff (van der Werff 1951) reported that fi nches eat the fruits of this species in the Galápagos.
Passifl ora tridactylites was described by J. D. Hooker in 1851. At the time he actually described what he considered to be three distinct species on the Galapagos Islands: P. lineariloba, P. tridactylites, and P. puberula. He based his descriptions primarily upon vegetative morphology. He described P. lineariloba as a slender vine having deeply trilobed leaves with long, very narrow lateral lobes that are broadly diverging. Hooker apparently did not see the fl owers of P. lineariloba because he does not describe them and the type specimen is sterile. Passifl ora tridactylites was described as having deeply trilobed leaves with subcordate bases and shorter, linear-oblong lateral lobes. Hooker described the fl owers of this species as large (3/4 inch in diameter), with fi ve linear, obtuse sepals with the ovary possessing a greatly elongated "pedicel" (androgynophore), and coronal fi laments that are subequal to the sepals. Passifl ora puberula was described as being covered in short, microscopic hairs and possessing trilobed leaves with cuneate bases and shorter, linear-lanceolate lateral lobes. Hooker goes on to describe the fl owers, which possess fi ve narrowly linear sepals that are pubescent, and fruits, which are ovate-oblong; though not mentioned in his description, the lectotype specimen of P. puberula possesses a very long androgynophore. Lawesson (1988) diff erentiated between P. suberosa and P. tridactylites, but did not list the synonyms of either species in his treatment. Hooker based his description of P. tridactylites on both vegetative and reproductive material with a detailed description of the fl ower and Lawesson (1988) used that name for the Galápagos entity, with P. lineariloba and P. puberula treated as synonyms. Th ough the type specimen of P. lineariloba is sterile, vegetatively identical specimens with very large fl owers and long androgynophores have been collected at the type locality. Th us, I have included it as a synonym of P. tridactylites rather than P. suberosa subsp. litoralis, which also occurs on the Galápagos Islands. Killip (1938) lumped P. lineariloba, P. tridactylites, and P. puberula with P. suberosa. He noted that the entities on the Galápagos Islands with very narrow leaf lobes that had been labeled P. lineariloba matched material collected by Saff ord and Mosier (227) from Florida. In addition, he noted that material similar to P. tridactylites exactly matched specimens collected by Brown (115) in Jamaica. Based upon vegetative characters alone he is quite correct, but the fl owers of these Galápagos specimens are distinctive. Th e specimens of Saff ord and Mosier and Brown are examples of P. pallida, and the fl owers and fruits of that species are far smaller than those of P. tridactylites. Lawesson (1988) diff erentiated between P. tridactylites and P. suberosa stating that the species were easily separated by the shape and size of the sepals and the androgynophore length.
Specimens examined. ECUADOR. Galápagos. Española: Española, Baur 160 (GH); "Gardner Island", Snodgrass & Heller 625 (GH); "Gardner Island", Snodgrass & Heller 321 (  Description. Slender, climbing, perennial vine 3 m long or more, densely pubescent with unicellular curved trichomes throughout (except ovary), 0.5-1.4 mm long, 0.02-0.06 mm wide, also sparsely, antrorsely appressed-puberulent with unicellular, curved trichomes on stems, leaves and stipules, 0.03-0.05 mm long, 0.02 mm wide. Flowering stems 0.7-2.2 mm in diameter, subterete to terete, with the base somewhat cork-covered. Stipules 4.1-8.5 mm long, 0.3-0.9 mm wide; petioles 0.7-1.9 cm long, narrowly ovate, acute to attenuate, longitudinally striate-nerved, eglandular (rare) or commonly bearing in the distal third (0.69-0.97 of the distance from the base toward the apex of the petiole) (1-)2, round or elliptic, opposite to alternate, long-stipitate, cupulate nectaries, 0.1-0.5 mm wide, 0.4-1.2 mm high. Laminas 3.5-8.5 cm long, 1.5-5.2 cm wide, unlobed to shallowly 3-lobed 0.05-0.72 of the distance to the leaf base, when present, lateral lobes 1.1-4.0 cm long, 0.5-3.0 cm wide, elliptic, acute to rounded, central lobes 3.5-8.5 cm long, 1.0-3.5 cm wide, ovate to elliptic, acute to attenuate, angle between the lateral lobes 53-115°, ratio of lateral to central lobe length 0.29-0.56, margins entire, primary veins 1(rare) or 3, diverging and branching at base, laminar nectaries absent; tendril 0.3-0.6 mm wide, present at fl owering node. Flowers borne in leaf axils. Pedicels 24.0-55.0 mm long, 0.3-0.8 mm wide; bract(s) absent or with one, narrowly ovate, acute bract, 0.9-1.8 mm long, 0.1-0.3 mm wide, the bract 20.6-34.8 mm from base of pedicel; spur(s) absent. Tubular fl owers 7.1-12.8 mm in diameter with stipe 2.9-7.4 mm long, 0.5-1.0 mm wide; hypanthium 7.1-12.8 mm in diameter; sepals 20.1-31.8 mm long, 3.4-6.9 mm wide, narrowly ovate, acute, abaxially and adaxially reddish purple (5RP 4/6-4/8) dried; coronal fi laments in 1 (rare) or 2 series, the outer 26-30, basally connate 1.1-3.8 mm, the free portions 5.8-10.3 mm long, 0.3-0.8 mm wide, linear to narrowly ovate, erect, reddish purple, lighter distally, ratio of coronal (fused and free portions) to sepal length 0.28-0.49, the inner not well-developed with 2-4 fi laments or well-developed (rare) with 30-31 fi laments, free or basally connate (rare) 0.8-0.2 mm, the free portions 1.1-2.9 mm long, 0.1-0.2 mm wide, linear, sometime capitellate, erect, appearing reddish purple when dried, ratio of inner coronal row to outer coronal row length (fused and free portions) 0.11-0.41; operculum 1.7-2.9 mm long, plicate, appearing light reddish purple dried, the margin with narrow minutely fi mbrillate teeth; nectary 0.09-0.13 mm high, 1.1-3.5 mm wide, sulcate; limen slightly recurved to erect, occasionally slightly inclined toward operculum, 0. Discussion. Passifl ora lancifolia is very similar to another Jamaican endemic, P. macfadyenii. Th ey both possess bright red, elongated tubular fl owers that are likely pollinated by hummingbirds. Th e two species can be easily separated utilizing both vegetative and reproductive characters. Passifl ora lancifolia possesses shallowly trilobed leaves (rarely unlobed) with the lateral lobes commonly signifi cantly less than half the length of the central lobe, and the central lobe is ovate and never narrowed at the base. Passifl ora macfadyenii possesses distinctly trilobed leaves with the lateral lobes commonly more than half the length of the central lobe, and the central lobe is obovate with a distinctly narrowed base similar to that in P. juliana and P. viridifl ora. Th e pedicels in P. lancifolia are greater than 2.3 cm long, whereas those of P. macfadyenii rarely exceed a length of 1.8 cm. Th e fl oral nectary of P. lancifolia is the widest in the supersection, greatly exceeding that of P. macfadyenii. Th e outer coronal fi laments are connate and often not adnate to the sepals or barely so in P. lancifolia, whereas those of P. macfadyenii are distinctly adnate to the sepals. Passifl ora lancifolia often has two rows of coronal fi laments (rarely with one row or a poorly developed inner row) and P. macfadyenii lacks an inner coronal row (or with a poorly developed second coronal row seen in one fl ower from a plant in cultivation, i.e., MacDougal 452 -cultivated from cuttings of Th omas 2032). Th e fruits of P. lancifolia and P. macfadyenii are distinct, with P. lancifolia having globose fruits and P. macfadyenii possessing fusiform fruits. Th e habitats of the species are also diff erent with P. lancifolia growing in tropical lower montane mist forests at 850-1220 m and P. macfadyenii found in tropical dry forests at 200-310 m.

Passifl ora lancifolia
Th e name P. lancifolia was originally published by Hamilton as "Passifl ora lancifolia Herb. Prof. Desv.," and the species has often been cited as "P. lancifolia Desv. in Ham." or "P. lancifolia Desv. ex Ham." However, in the preface of his book, it appears that Hamilton himself took responsibility for the new species and genera described therein and only acknowledged the advice and assistance of Desvaux (see MacDougal and McVaugh 2001 for further details). Soon afterwards, Don (1834) described the taxon P. lanceolata. However, Don's description of P. lanceolata is identical to that of P. lancifolia in Hamilton and is based upon the same type material, therefore, the name P. lanceolata G.Don is a nomenclatural synonym of P. lancifolia Ham. In 1850, Macfadyen wrote his second volume of Flora of Jamaica and included in it the description of a diff erent plant, which he called P. regalis, now known as P. macfadyenii C. D. Adams. However, Macfadyen unexpectedly passed away before the publication of his fl ora, though it was distributed. As a result, several authors viewed the new species that were described by Macfadyen as ineff ectively published and began to publish new species based upon his work. Grisebach (1860) was one of these authors and published a description of P. regalis, which he attributed to Macfadyen. However, the species that he described was P. lancifolia and not Macfadyen's P. regalis. In addition, Ramírez Goyena (1909) published a description of P. regalis, which he attributed to Macfadyen, but the species that he described was also P. lancifolia and a later homonym of P. regalis Macf. ex Griseb. Incidentally, Ramírez Goyena's description of P. regalis, other than being in Spanish and not in English, is virtually identical to that of Grisebach. Killip (1938) placed P. lancifolia together with P. viridifl ora in the subgenus Chloropathanthus. However, the discovery of P. juliana, a species that very closely resembles P. viridifl ora but is clearly a member of supersection Cieca, reinforced MacDougal's hypothesis (1983) that the apetalous, tubular-fl owered species (including P. lancifolia) belong in supersection Cieca (MacDougal 1983(MacDougal , 1992. Benson et al. (1975), in a study of the coevolution of plants and herbivores, reported that Dryas julia is an herbivore of P. lancifolia.
Phenology. Flowering and fruiting from December to February, sometimes fl owering in June.
Distribution. Endemic to Jamaica, in the parishes of St. Andrew and St. Th omas. Tropical dry forests in roadside thickets and wooded limestone hills near Lucky Valley (St. Andrew) and Cambridge Hill (St. Th omas); growing on shrubs, small trees, limestone boulders and rocks on very limited to moderately developed soils; ca. 200-310 m.
Discussion. As mentioned under P. lancifolia, P. macfadyenii is somewhat similar to that taxon but diff ers from it in characters of the leaf, fl ower, and fruit. Both species are quite distinct and can be easily separated in the fi eld and herbarium. It is interesting that the leaf shape of P. macfadyenii is very similar to that of P. juliana and P. viridifl ora. Killip (1938), under his description of P. lancifolia, also noticed their vegetative similarities.
Passifl ora macfadyenii is very restricted in its distribution and has only been collected in the vicinity of Lucky Valley in the dry tropical forests of the Port Royal Mountains, St. Andrew, Jamaica. I visited this area in June of 2000, but the region was experiencing a severe drought and four days of searching for the plant revealed neither vegetative nor reproductive material. Elma Kay (St. Louis University and Missouri Botanical Garden) and George Proctor (University of the West Indies and the Institute of Jamaica) have also made several trips to the area and have not been able to fi nd P. macfadyenii. It was last collected in 1979 (Th omas 2032, 2034) and was listed as a rare plant in the 1997 IUCN Red List of Th reatened Plants. It is my opinion that its status should be upgraded to extinct/endangered. It is fortunate that MacDougal obtained cuttings of P. macfadyenii from Th omas (MacDougal 452, Th omas 2032) and grew the plant in the greenhouses at Duke University from 1979-1982; it is no longer in cultivation. Th anks to their eff orts we have a better understanding of the biology of this very rare taxon.
In an unpublished manuscript, MacDougal determined the total sugar concentration measured as sucrose equivalents in percent weight per total weight to be 29-44% in P. macfadyenii. He also found the fl ower to have no odor. Th e fl ower shape and morphology, combined with these data, indicate that P. macfadyenii is (or was) likely utilized by hummingbirds.
Phenology. Flowering and fruiting March to December. Distribution. Northern Mexico and southern Texas in the United States. Arid and semiarid thorn scrub (e.g., Mesquite-Black brush, Opuntia-Prosopis scrub, Tamaulipan thorn scrub) and grasslands; climbing on shrubs or scrambling on limestone outcrops and hills, or in open grassy areas on very limited to moderately developed soils; ca. 150-1500 m.
Discussion. Passifl ora tenuiloba is very distinctive in the form of the leaves. It possesses leaves that are shallowly to deeply 3-to 5-lobed, often with lateral lobes that are up to 8.0 cm long and between 0.2 and 2.1 cm wide. Th e lateral lobes frequently possess 2-3 lobes at their apices. Th e central lobe is short (<1.0 cm) or longer (to 3.7 cm), sometimes with three lobes at its apex. Th e petiolar glands are positioned on the distal half of the petiole, often at the petiole apex or even on the base of the leaf. Passifl ora tenuiloba also has very distinctive seeds with reticulate centers and grooved edges.
Passifl ora tenuiloba occurs in southwest Texas and northern Mexico along with P. pallida. Th e small fl owers of these two species are somewhat similar, but they can be easily separated by vegetative characters. Th e most obvious diff erence is the shape of the lamina, with P. tenuiloba possessing leaves that are transversely elliptic and P. pallida possessing leaves that are ovate to elliptic in general outline. In addition, the fl owers of P. tenuiloba have a wider hypanthium than those of P. pallida and have more and commonly longer fi laments in their coronal rows. Th e seeds of P. tenuiloba are 4.1-5.8 mm long, whereas those of P. pallida do not exceed a length of 3.5 mm.
Passifl ora tenuiloba has been included in three other studies of passionfl owers. Benson et al. (1975) found that Agraulis vanillae (Gulf Fritillary) is an herbivore of this species. Klucking (1992) found that the leaf venation pattern of this species is similar to P. sexocellata and P. eglandulosa and was classifi ed as actinodromous and pinnate secondary venation with irregular to regular intercostal venation consisting of lineate and transverse veins. According to Klucking, the leaves of P. tenuiloba are more like those of P. eglandulosa, because they have acute lateral lobes, an angle between the lateral veins that is between 120 and 140°, and leaf bases that are cordate.
Phenology. Flowering and fruiting January-May, July-September and December. Distribution. El Salvador, Guatemala, and Honduras. Growing in shrubs and small trees in shady ravines and at the edges of premontane to montane broad-leaved forests on volcanic cones; 1500-2650 m.
Discussion. For many years after the publication of Killip's 1938 monograph, the name P. trinifolia Mast. was applied to two distinct taxa: P. eglandulosa and P. trinifolia. In fact, Standley and Williams (1961), in their description of P. trinifolia, combined information from Killip's description of P. trinifolia, which strictly applied to P. trinifolia in the sense of Masters, and their own personal observations of P. eglandulosa (MacDougal 1988). It is true that the two species both possess wide foliose stipules, similarly trilobed leaves at fertile nodes and seeds with the micropylar end and chalazal beak erect and not inclined toward the raphe. However, P. eglandulosa is distinguished by fl owers with longer fl ower pedicels, spurs that occur between each of the sepals, narrower sepals, narrower outer coronal fi laments, shorter inner coronal fi laments that are not broadly capitate, narrow limen fl oors, short staminal fi laments, and anthers that present pollen laterally as opposed to subproximally. Th e seeds are longer and wider than those of P. trinifolia, and as its name implies, P. eglandulosa lacks both laminar and petiolar nectaries; petiolar nectaries have been seen on only one specimen, M. Veliz 16059. Passifl ora eglandulosa possesses fl ower buds that are slightly horned at the apex and fl owers that are oriented above rather than near or below the horizontal plane. In comparing the habitats of the two species, MacDougal found that P. eglandulosa is found in shady ravines and at the edges of wet premontane to montane broad-leaved forests on volcanic cones, whereas P. trinifolia is found in open, seasonally dry pine/oak forests on rock outcrops. In addition, P. eglandulosa is a larger plant that may climb to 4 m or more, but P. trinifolia rarely exceeds a height of 1 m. Th e chartaceous leaves of P. eglandulosa are bright green adaxially and possess drip tips, but the leaves of P. trinifolia are dark green, lack long drip tips and are very stiff and rigid (MacDougal 1988).
Passifl ora eglandulosa is also similar vegetatively to P. tacanensis, a species found in montane forests on Volcán Tacaná of Chiapas, Mexico. Both species possess wide, foliose stipules. However, the two species are easily separated because P. tacanensis possesses petiolar nectaries. Th e fruits of P. eglandulosa also possess fewer than 10 seeds, whereas P. tacanensis possesses ca. 20 seeds per fruit.
Phenology. Th e species has been collected in fl ower in April. Distribution. Endemic to Guatemala in the department of Alta Verapaz at ca. 762 m altitude. Based upon locality information included on the herbarium specimen and information gathered by J. M. MacDougal (pers. comm.) on a recent trip to the type locality, P. clypeophylla is (or was) likely found on slopes of premontane tropical moist forest.
Discussion. Passifl ora clypeophylla is known only from the type collection from Alta Verapaz, Guatemala. Passifl ora clypeophylla is distinctive in supersection Cieca because of its large, conspicuously peltate leaves that are deltoid in general outline. Th e fl owers are not known to be borne in infl orescences and the pedicels are greater than 16.8 mm long. Th e fl oral stipe of P. clypeophylla is also one of the longest in the supersection and is greater than 9.4 mm long. In addition, the plant has very shallow leaf lobes (0.03-0.07 of the distance from the leaf outline to the leaf base).
Passifl ora clypeophylla resembles both P. trinifolia and P. sexocellata, which are somewhat similar vegetatively and also occur in Guatemala. Passifl ora clypeophylla is easily distinguished from P. trinifolia by its considerably narrower stipules, the obtuse to rounded leaf lobes that are very shallow and the leaves that are coriaceous as opposed to chartaceous in texture. Th e primary diff erence between Passifl ora sexocellata and P. clypeophylla is the ratio of the lateral to central lobe length. Passifl ora clypeophylla has lateral and central leaf lobes that are nearly equal in length, whereas P. sexocellata has lateral lobes that are commonly 1.3 to 2.8 times longer than the central lobes. Passifl ora sexocellata also commonly has a shorter central leaf lobe and more laminar nectaries than P. clypeophylla. As with P. trinifolia, the lateral leaf lobes in P. sexocellata are commonly acute as opposed to obtuse to rounded. Th e one known fl ower of P. clypeophylla has fewer fi laments in the outer coronal row (28 fi laments) than either P. trinifolia (35-39 fi laments) or P. sexocellata (40)(41)(42)(43)(44)(45)(46)(47)(48)(49)(50). Th e staminal fi laments in P. clypeophylla are nearly equal to the androgynophore length, but the fi laments in P. sexocellata are commonly half the length of the androgynophore.
Th e seedling leaves of several species in supersection Cieca (e.g., P. sexocellata, P. megacoriacea, P. juliana, and P. viridifl ora) are peltate and very similar in shape to the mature leaves of P. clypeophylla, and evolution by neoteny in this taxon seems plausible.
Th ere are only two known specimens of P. clypeophylla in the world, one at the Kew Herbarium and the other at the United States National Herbarium. In his description of P. clypeophylla Masters did not cite a herbarium, only a collection. Th e specimen at K is much better than the one at US, so I have designated it the lectotype.

Passifl ora obtusifolia
Phenology. Flowering and fruiting October to January and May. Distribution. Costa Rica, El Salvador , and Mexico. Tropical deciduous and subdeciduous forests or disturbed areas in the Pacifi c lowlands and foothills; near sea level to 300 m in Mexico, 650-1200 m in El Salvador and Costa Rica.
Discussion. As noted by MacDougal and McVaugh (2001), Passifl ora obtusifolia is quite variable in its vegetative morphology, especially in the depth of the leaf lobes, the shape of the lobe apices, and the number of laminar nectaries. Despite its name, the lobes of P. obtusifolia are commonly acute. Th e type is an illustration of an unusual form that has only been collected again near Cerro de Ortega, Colima, Mexico (Lott 840), not far from the type locality. Th e illustration shows a plant with shallowly trilobed leaves with obtuse lateral lobes, rounded to emarginate central lobes, and six laminar nectaries per leaf, with two glands situated proximal to the lateral leaf veins.
Passifl ora obtusifolia is similar to Passifl ora mcvaughiana and both are found in southwestern Mexico. However, these species diff er in leaf shape, depth of lobing, number of laminar nectaries, number of petiolar nectaries, pedicel length, sepal length, outer coronal length and shape, seed size, and habitat. Passifl ora obtusifolia can also be found in locations somewhat near P. eglandulosa. At fi rst glance these two species are somewhat similar vegetatively with their distinctly trilobed leaves. However, the stipules of P. eglandulosa are much wider and foliose, the leaf bases are cordate, and the leaf apices are acuminate. In addition, P. eglandulosa does not possess infl orescences and its fl owers are more delicate with narrower sepals and thinner outer coronal fi laments. Passifl ora obtusifolia is also similar to P. suberosa subsp. litoralis. However, P. suberosa subsp. litoralis is never peltate at the reproductive nodes, whereas P. obtusifolia is commonly peltate. Passifl ora suberosa subsp. litoralis does not produce fl owers in long infl orescences. Th e fruits of P. obtusifolia are over 20 mm long and 18 mm wide, but the fruits of P. suberosa subsp. litoralis rarely exceed a length of 12 mm and a width of 10 mm.
Phenology. Flowering and fruiting August to November. Distribution. Mexico, in the Pacifi c lowlands and foothills of Jalisco, Colima, and northern Michoacán. Disturbed tropical deciduous or semideciduous low and medium forests (selva baja caducifolia and selva mediana subcaducifolia); growing on shrubs, trees, boulders, and rocks (sometimes limestone); sea-level to ca. 610 m.
Discussion. Passifl ora juliana is most closely related to P. viridifl ora and aside from fl oral adaptations in P. viridifl ora resulting from a shift in pollinators, these two species with greenish yellow fl owers borne in conspicuous, indeterminate, terminal infl orescences are very similar. Both species possess large, peltate, trilobed leaves that have a central lobe that is distinctly narrowed at the base. Th ey both may possess stems that have some red pigmentation, but those of P. viridifl ora are generally bright red, while those of P. juliana are commonly reddish purple. Passifl ora juliana can also be separated from P. viridifl ora vegetatively because that species has small, narrowly ovate stipules, as opposed to the larger, ovate, foliose stipules of P. juliana. Passifl ora juliana bears the shallow cup-shaped fl owers typical of most of the members of the supersection and subgenus, whereas P. viridifl ora possesses long, tubular fl owers with a greatly elongated androgynophore. Passifl ora juliana is a very distinctive taxon possessing the shortest fl oral stipe in supersection Cieca, a limen fl oor that is distinctly purple and an androgynophore fl ushed with purple at the base or to just above the middle.

Passifl ora viridifl ora
Phenology. Flowering and fruiting throughout the year. Distribution. Mexico, in the Pacifi c lowlands and foothills of southern Michoacán, Guerrero and Oaxaca. Disturbed tropical deciduous or semideciduous low and medium forests (selva baja caducifolia and selva mediana subcaducifolia); growing on shrubs, small trees, boulders and rocks (sometimes limestone) on very limited to moderately developed soils; sea-level to ca. 610 m.
Discussion. Vegetatively, P. viridifl ora and P. juliana are very similar, and the most obvious diff erence between them is the size and shape of their stipules. However, P. viridifl ora also diff ers from P. juliana in its adaptations for hummingbird pollination including: vegetative parts that are commonly accentuated with or entirely bright red, a greatly elongated androgynophore that far exceeds the length of the stamen fi laments, no inner coronal fi laments, a very narrow limen fl oor, wide fl oral nectary, long operculum that is not incurved at the margin but erect and lays against the androgynophore, fused sepals that are greatly elongated, pollen that is presented laterally, and a sulcate fl oral nectary fl oor.
On an herbarium specimen collected by W.L. Forment (1125), he indicated that P. viridifl ora is utilized by hummingbirds, which is consistent with its fl oral morphology and lack of fl oral fragrance.
Passifl ora viridifl ora has been placed at various generic (e.g., Murucuia) and infrageneric levels (e.g., subg. Chloropathanthus) within the family Passifl oraceae. Th e elongated, tubular fl owers of this taxon inspired many previous workers to group it with other taxa that possess tubular fl owers or in a group of its own because the fl owers are not only tubular but also apetalous. Killip (1938) placed it in the subgenus Chloropathanthus with P. lancifolia, an apetalous Jamaican endemic. MacDougal (1983) was the fi rst to suggest that P. viridifl ora be placed within Cieca based upon its apetalous fl owers and fl avonoid chemistry. In 1992, MacDougal resolved the placement of P. viridifl ora by describing P. juliana, a species clearly referable to Cieca and morphologically similar to P. viridifl ora. Both the molecular and morphological data in this study also show that P. juliana and P. viridifl ora are sister species.
Phenology. Flowering and fruiting June to December. Distribution. Mexico, in the states of Jalisco, Mexico, and Guerrero. Pine and oak forests (bosque de pino y encino) or montane mesophytic forests (bosque mesófi lo de montaña); growing in trees and on the steep banks of canals (barrancas) or streams, and moist hillsides; 1100-2000 m.
Discussion. Passifl ora mcvaughiana is one of four species found in Mexico previously known under the name of P. coriacea Juss. Th e other two species are Passifl ora obtusifolia and P. tacanensis, which are both extremely similar vegetatively to P. mcvaughiana. Passifl ora mcvaughiana can usually be separated from P. obtusifolia because P. mcvaughiana commonly has a central leaf lobe that is nearly as long as the lateral lobes at fertile nodes, as opposed to having lateral lobes that are commonly twice as long as the central lobe in P. mcvaughiana. Passifl ora obtusifolia is commonly 3-lobed more than 0.20 the distance to the base, as opposed to 3-lobed less than 0.20 the distance to the base in P. mcvaughiana. Passifl ora mcvaughiana lacks laminar nectaries, whereas P. obtusifolia commonly has 2-6 nectaries present between the primary leaf veins. Flowers are rarely produced in infl orescences in P. mcvaughiana, but P. obtusifolia commonly has very long infl orescences. Th e pedicel in P. mcvaughiana is longer than 10 mm, but the pedicel in P. obtusifolia is commonly less than 10 mm long. Th e outer coronal fi laments of P. mcvaughiana are longer than 4.0 mm, linear, and dull purple toward their bases, those of P. obtusifolia are commonly less than 4.0 mm long, linear/capitellate and greenish yellow or greenish yellow with a fl ush of reddish purple at the base. Passifl ora mcvaughiana possesses the widest seeds in the supersection (over 3.6 mm wide) and P. obtusifolia has seeds that are less than 2.3 mm wide. Additionally, P. mcvaughiana and P. obtusifolia occupy diff erent habitats, with P. obtusifolia commonly occurring in lower elevation tropical deciduous or semideciduous forests in Pacifi c lowlands and foothills and P. mcvaughiana in high elevation oak, pine/oak, pine or montane mesophytic forests of Mexico.
Passifl ora tacanensis is a newly discovered species from Volcán Tacaná, Chiapas, Mexico. Like P. mcvaughiana, it occurs in montane mesophytic forests. However, P. tacanensis is easily separated from P. mcvaughiana by its foliose stipules that are more than 3 mm wide. Th e fruits of P. tacanensis also possess ca. 20 seeds, whereas those of P. mcvaughiana produce only 2-10 seeds.
Passifl ora mcvaughiana is also quite similar vegetatively to P. sexocellata, though this species does not occur in southwestern Mexico. Th e leaves of P. mcvaughiana are not as coriaceous as those of P. sexocellata and are darker green. In addition, P. sexocellata always possesses 4-13 laminar nectaries while P. mcvaughiana has none. Th e petiolar nectaries of P. sexocellata are commonly positioned on the proximal half of the petiole, whereas those of P. mcvaughiana are positioned on the distal half of the petiole. Flowers are often produced in long infl orescences in P. sexocellata, and P. mcvaughiana commonly lacks infl orescences. Th e fruits of P. sexocellata are also much larger than those of P. mcvaughiana and possess between 40 and 50 seeds per fruit.
Discussion. Passifl ora tacanensis is known only from the general type locality and though Martínez (Martínez 20782) states that the fl owers are purple, the specimen does not possess fl owers and I have not been able to locate any duplicates. It was found in montane mesophytic forests on Volcán Tacaná in southwestern Mexico and was collected in May during the rainy season.
Phenology. Flowering and fruiting throughout the year. Distribution. Bolivia, Colombia, Ecuador, Perú , and Venezuela; reported once from Guyana (Lejos 43, B, destroyed). Growing in shrubs and small trees in secondary successional areas, along the edges of moist tropical forests near rivers and streams, and along the seashore, 0-1500 m.
Ethnobotany. Timothy Plowman in a note on a specimen collected by him in 1976 (T. Plowman 6029), noted that in Perú a medicine for the liver is prepared from P. coriacea by boiling the whole plant and then drinking the syrup.
Discussion. Passifl ora coriacea is extremely similar to P. sexocellata and P. megacoriacea in its vegetative morphology, but is easily distinguished by its fl owers. Th e fl owers of P. coriacea possess long fl oral stipes as compared to their pedicels (the stipes are usually two to three times the length of the pedicels) and an operculum that is dark reddish purple. Passifl ora sexocellata has fl oral stipes that are commonly shorter than or equal in length to the pedicels and an operculum that is greenish yellow with a fl ush of dark reddish purple at the base and a white margin. Passifl ora megacoriacea possesses fl oral stipes that are commonly less than half the length of the pedicels and an operculum that is greenish yellow with a white margin or greenish yellow with a mere fl ush of reddish purple at the base and a white margin. Passifl ora coriacea is also distinguished by outer coronal fi laments that may appear banded with light to dark reddish purple. In addition, the outer coronal fi laments are more dilated distally, much like P. megacoriacea but in contrast to P. sexocellata. Th e limen fl oor in P. coriacea is very light greenish yellow with dark reddish purple spots and streaks, again much like P. megacoriacea. Passifl ora sexocellata usually possesses a very dark red limen fl oor. Schlechtendal (1854) attempted to use mostly vegetative characters to distinguish P. coriacea from P. sexocellata; incidentally, he was the fi rst to notice diff erences in the stipe and pedicel lengths of the two species. He used the following characters to differentiate them: position of the petiolar nectaries, the number of laminar nectaries, the shape of the stem, leaf venation, the leaf margin, and the leaf texture. However, in my analysis of these species, I did not fi nd any of these vegetative characters to be wholly reliable in distinguishing between these two species. Both have petiolar nectaries that occur in various positions below the middle of the petiole, stems that are terete to somewhat compressed, fi ve distinct leaf veins, thick leaf margins and coriaceous leaves. Passifl ora sexocellata does tend to have fewer nectaries than P. coriacea on average, but there is a signifi cant amount of overlap in the range of variation.
A clone of P. coriacea (MacDougal 3029) did not produce fruits by autogamy in years in cultivation. Th is greenhouse accession was given to me by MacDougal, who originally received it as seedlings from J. Zarucchi (Zarucchi et al. 6102).
Heliconius erato (Lepidoptera: Nymphalidae, Heliconiinae) has been reported to be an herbivore of P. coriacea in the central Colombian valleys (Cauca and Magdalena) (Benson et al. 1975). Fajardo et al. (1998) in a study on the genetic variation analysis of the genus Passifl ora using RAPD markers, used P. coriacea and P. adenopoda DC. as representatives of taxa from subgenus Decaloba. Th ey found P. coriacea to be genetically distant from the other taxa in his study, including P. adenopoda, but due to insuffi cient data, they were not able to discuss the signifi cance of this result (Fajardo et al. 1998).
In Antoine Laurent de Jussieu's original description of P. coriacea (1805) he included a detailed diagnosis and drawing of the species. Th e lectotype of P. coriacea (at P), closely resembles the drawing in Jussieu, but there are no locality data on the specimen. Th e isolectotype of P. coriacea consists of two leaves and a small portion of the stem and does not resemble the type drawing of the species, but written on the specimen are locality and descriptive data in Jussieu's hand. Santander: River Suratá valley, between Bucaramanga and El Jaboncillo, 800-1500 m, Killip & Smith 19062 (GH, US) (F, MO). San Martín: Prov. Mariscal Caceres, Dtto. Tocache Nuevo, Schunke 3823 (F, GH, MO, US).
Phenology. Flowering and fruiting throughout the year. Distribution. Colombia, Costa Rica, and Panama. Growing in shrubs or trees in secondary successional areas, along the edges of tropical moist to premontane wet forests, and near the seashore, 0-1100 m altitude.
Discussion. Passifl ora megacoriacea is relatively common in Costa Rica and Panama. John MacDougal brought my attention to the variation of vegetative and fl oral characters of some of the Costa Rican and Panamanian specimens then identifi ed as P. coriacea.
Passifl ora megacoriacea, as noted above in the discussion of P. coriacea, is very similar to P. coriacea and P. sexocellata, and although not sympatric, without reproductive material it can be diffi cult to separate them. Passifl ora megacoriacea may be recognized by commonly having petiolar nectaries found on the distal half of the petiole, (0.30-) 0.50-0.77 of the distance from the base toward the apex of the petiole, and although that overlaps the 0.21-0.54(-0.64) range of P. coriacea and P. sexocellata, the character is easily seen in herbarium specimens. Passifl ora megacoriacea can also possess deeply trilobed leaves (commonly 0.11-0.61 the distance to the base), especially in populations along the Pacifi c coast of Costa Rica and in the Panamá Canal Zone, whereas P. coriacea and P. sexocellata do not possess deeply trilobed leaves (commonly less than 0.11 the distance to the base). Th e reproductive structures of these three species provide a number of distinguishing characters. Passifl ora megacoriacea possesses fl oral stipes that are commonly less than half the length of the pedicels, whereas P. coriacea possess stipes that are usually two to three times the length of the pedicels and Passifl ora sexocellata has fl oral stipes that are commonly just shorter than or rarely up to two times the length of the pedicels. Th e overall size of the fl ower of P. megacoriacea exceeds that of both P. coriacea or P. sexocellata, with P. megacoriacea commonly having a wider hypanthium, longer sepals, larger and fewer outer coronal fi laments, a longer androgynophore, longer staminal fi laments, longer anthers, and a longer operculum. Th e most informative of these is the length of the androgynophore, with P. megacoriacea having an androgynophore that is 6.9-8.8 mm long and the androgynophores of both P. coriacea and P. sexocellata not exceeding a length of 5.9 mm. In addition, the nectary fl oor is raised in P. megacoriacea, never raised in P. sexocellata, and only rarely raised in P. coriacea. Th e outer coronal fi laments of both P. megacoriacea and P. sexocellata are erect, while those of P. coriacea spread to ca. 140-160°. Th e fl owers of P. megacoriacea are commonly referred to as white, greenish white, or cream on herbarium labels and this is due to it having no (or relatively little) reddish purple coloration in the mature fl owers; the fl owers of P. coriacea and P. sexocellata both commonly have a signifi cant amount of reddish purple coloration.  (MO). San Blas: on mainland in front of Ustupo, D'Arcy 9527 (MO).

Passifl ora sexocellata
Phenology. Flowering and fruiting throughout the year. Distribution. Mexico and Central America (except Costa Rica and Panama). Growing in shrubs, trees or trailing on the ground in secondary successional areas, along the edges of semideciduous to deciduous, dry to wet tropical forests, both inland and near the seashore, 0-1171 m.
Ethnobotany. Th e vine is sold in Guatemalan herb markets and is sold dried where the plant does not grow naturally (Morton 1981). A decoction of the leaves is commonly taken as a diuretic, especially in the treatment of kidney infections (Morton 1981). In El Salvador and Honduras the leaves are combined with lard and used as a poultice on wounds and swellings (Morton 1981).
Discussion. Passifl ora sexocellata is very similar to P. coriacea and P. megacoriacea, and some of their similarities and diff erences are discussed under their respective descriptions. According to Jan Meerman (pers. comm.), Passifl ora sexocellata and P. xiikzodz grow side by side in Belize, with P. sexocellata growing in the sun and P. xiikzodz growing in the shade. Where these two species are found in the Yucatán Peninsula of Mexico, I found that P. sexocellata occurs in wetter forests along rivers and lakes to the west and P. xiikzodz and the related P. itzensis are found in drier forests to the east. However, MacDougal concluded that these plants grow together at some sites in the Yucatán (MacDougal 1992). Th ese two species are easily separated because P. xiikzodz and P. itzensis possess petiolar nectaries at or near the apex of the petiole whereas P. sexocellata has petiolar nectaries on the proximal half of the petiole. In addition, numerous fl oral characters can be used to distinguish between them. Th e most obvious diff erence is the number of coronal rows, with P. xiikzodz and P. itzensis possessing seven series and P. sexocellata possessing only two. Klucking (1992) classifi ed the leaf venation of P. sexocellata, identifi ed as P. coriacea, as actinodromous and pinnate secondary venation with irregular to regular intercostal venation consisting of lineate to transverse veins. Th e peltate, trilobed leaf illustrated is the typical form for P. sexocellata. Th ere are three primary veins and two acrodromal veins which extend two-thirds the length of the lateral lobes, the lateral lobes are acute, the angle between the lateral veins is 150°, and there are six laminar nectaries apparent on the abaxial surface (Klucking 1992).
In Belize, Meerman (2001) found that Heliconius erato is an herbivore of P. sexocellata (which he identifi ed as P. coriacea). Benson et al. (1975) found that Dryas julia and Heliconius erato were herbivores of P. sexocellata (again, identifi ed as P. coriacea).
In 1990, Joanna Turner collected Plaster bees, Colletes sp., that regularly visited fl owers of P. sexocellata in Belize. Th e bees are approximately 10 mm long, 3-4 mm high, including some off -the-ground leg clearance, and have a thorax that is 2.0-2.5 mm high, and were identifi ed by Rick Clinebell at MO (pers. comm.).
Passifl ora sexocellata was originally described by Schlechtendal in 1854. He cited "P. marmorea hort.", as a synonym, but this horticultural name was not validly published. It is interesting that the specifi c epithet "marmorea" means marbled, as the leaves of this species are often variegated. Holm-Nielson et al. (1988), in the synonomy of P. coriacea, stated that P. sexocellata is an illegitimate name that was based upon material of P. coriacea Juss. and P. diff ormis Kunth. However, I do not see any reason why Schlechtendal's species has to be considered illegitimate. He carefully describes the plant from cultivated material that he had at hand in the Botanical Gardens in Halle, Germany and spends a paragraph diff erentiating his species from both P. coriacea and P. diff ormis. I was unable to locate the type of P. sexocellata, and U. Braun (curator of the herbarium at the Herbarium at the Institut fur Geobotanik und Botanischer Garten, Halle) was unable to fi nd any material under the name P. sexocellata. Braun was also unable to locate appropriate material under P. coriacea or P. diff ormis. Other species of Passifl ora from Mexico and Central America were in cultivation in Europe by 1830 (Loudon 1830), and it is plausible that Schlechtendal had such material at hand when he described P. sexocellata. Schlechtendal's P. sexocellata seems to fi t the description of the Mesoamerican entity that I am recognizing as a species distinct from P. coriacea and other similar taxa from supersection Cieca. However, some of the vegetative characters that he uses to distinguish P. sexocellata are actually quite variable, but he only had one live specimen available to him when he described the species. He describes the fl ower as having fi ve green sepals, outer coronal fi laments that are "lilac" at the base but "greenish yellow" otherwise, inner coronal fi laments that are dilated at the apex and "lilac" in color at the tips and lighter toward the base, an operculum that is dull "lilac" at the base and becoming "greenish yellow" toward the apex, and a "greenish yellow" androgynophore. Th e use of the term "lilac" is somewhat misleading, but the description of how the colors vary on the various parts of the fl ower is diagnostic. For example, P. coriacea possesses outer coronal fi laments that are reddish purple at the base but obviously white toward the tips with a band of reddish purple and not "greenish yellow." In addition, the operculum of P. coriacea is wholly reddish purple. However, P. sexocellata possesses outer coronal fi laments that are reddish purple at their bases, greenish yellow at their middles and yellow at their apices and an operculum that is dark reddish purple at the base and greenish yellow otherwise (often with a white margin). Based upon Schlechtendal's detailed description, I apply the name P. sexocellata to this species and have designated a neotype that perfectly illustrates the diagnostic characters of the taxon, with the colors of the corona and limen Type. Based on Passifl ora xiikzodz J.M. MacDougal subsp. itzensis J.M. MacDougal Description. Slender, low-climbing or trailing, perennial vine 1-3 m or more, minutely antrorsely appressed-puberulent throughout with unicellular, curved trichomes, 0.06-0.11 mm long, 0.02 mm wide. Flowering stems 1.4-2.3 mm in diameter, terete or somewhat compressed, greenish yellow (5GY 8/4) to very dark reddish purple (5RP 2.5/2). Stipules 2.5-5.6 mm long, 0.4-0.6 mm wide, narrowly ovate, acute to slightly attenuate, longitudinally striate-nerved; petioles 0.9-1.8(-3.0) cm long, inserted 2.4-6.1(-7.0) mm from the basal margins of the peltate blades, with 2, round or elliptic, opposite, sessile, discoid nectaries with fl at rims, 1.3-1.9 mm wide (on the widest axis), 0.5-0.9 mm high, borne in the distal third of the petiole (0.62-0.83 of the distance from the base toward the apex of the petiole). Laminas 2.3-4.6 cm long, 5.0-12.4(-13.1) cm wide, coriaceous, often variegated along primary veins and major secondary veins, ratio of leaf width to central vein length measured from point of petiole insertion 1.9-5.1, depressed obovate to transversely elliptic (widely divaricately bilobed), lateral lobes (3.5-)4.3-7.4 cm long, 1.7-4.1 cm wide, elliptic, acute to slightly attenuate, central lobe commonly obsolete or present as an obtuse tip, central vein 1.8-3.1(-4.1) cm long (measured from point of petiole insertion), angle between the lateral lobes (85-)103-140°, ratio of lateral lobe to central vein length 1.4-2.8, margins entire, hyaline, primary veins 3, diverging and branching above base, laminar nectaries present, 6-19, submarginal, associated with the minor veins of the abaxial surface, 0.6-1.8 mm in diameter, widely elliptic to circular, sessile; tendril 0.4-0.9 mm wide, present at fl owering node, absent in infl orescence. Flowers borne in leaf axils or infl orescences; infl orescences 5.3-9.6 cm long, associated reduced laminas 1.9-2.5 mm long, 1.3-2.7 mm wide. Pedicels 1. 3-3.4(-5.8) mm long, 0.6-1.1 mm wide, (1-)2 per node; bract(s) absent; spur(s) absent. Flowers 20.3-25.5 mm in diameter with stipe 9.1-14.3 mm long, 0.6-1.0 mm wide; hypanthium 4.0-6.2 mm in diameter; sepals 7.5-9.8 mm long, 2.6-4.3 mm wide, ovate-triangular, acute, abaxially and adaxially greenish yellow or sometimes greenish yellow with very dark reddish purple streaks abaxially; coronal fi laments in 7 series, the outer 22-31, 6.3-8.1 mm long, 0.2-0.3 mm wide, linear, spreading fl at, the tips often slightly incurved, very dark reddish purple (5RP 2.5/2-3/2) with yellow (5Y 8/4-8/6) at tips, ratio of outer coronal row to sepal length 0.67-0.97, the second 20-30, 2.5-5.0 mm long, 0.1-0.2 mm wide, linear, spreading fl at, very dark reddish purple with yellow tips, ratio of second coronal row to outer coronal row length 0.33-0.64(-0.75), the third ca. 50, 0.7-2.1 mm long, 0.05-0.13 mm wide, linear, spreading fl at, very dark reddish purple with yellow tips, ratio of third coronal row to second coronal row length 0.22-0.59, the fourth through seventh ca. 100 per series, 0.7-1.1 mm long, 0.05-0.11 mm wide, linear, capitate, erect, very dark reddish purple, ratio of coronal rows 4-7 to third coronal row length 0.51-0.62(-0.90); operculum 0.3-0.4 mm long, denticulate, very dark reddish purple, nectary absent; limen absent, limen fl oor 2. 8-4.1(-5.7) mm in diameter, very dark reddish purple; androgynophore appearing absent, or 0.3-1.7 mm long, 0.9-1.8 mm wide; free portions of the staminal fi laments 1.9-3.4 mm long, 0.5-0.8 mm wide, linear, very dark reddish purple; anthers 1.3-2.0 mm long, 0.7-1.4 mm wide, introrse at anthesis with their axes maintained more or less parallel to the fi lament, anthers dehiscing distally; styles 1.8-3.1 mm long including stigmas, 0.3-0.5 mm wide, very dark reddish purple or greenish yellow with very dark reddish purple tinge toward base; stigmas 0.9-1.4 mm in diameter; ovary 1.7-2.4 mm long, 1.2-1.3 mm wide, widely ellipsoid to globose, greenish yellow. Berry 26.0 mm long, 14.0 mm in diameter, ovoid to obovoid, greenish yellow with white spots, becoming soft at the base at maturity. Seeds 30-40, 5.0-5.5 mm long, 2.0-2.2 mm wide, 1.3-1.8 mm thick, elliptic to slightly obovate in outline, acute at both ends, reticulate-foveate with each face marked with 20-22 foveae. Germination type epigeal.
Phenology. Flowering and fruiting September to June. Distribution. Mexico, in the states of Campeche, Quintana Roo, and Yucatán. Tropical semideciduous forests (selva mediana subcaducifolia and selva mediana subperennifolia); growing in shrubs or trailing along the ground on soil of little depth, lying directly on top of limestone; 0-23 m.
Discussion. In 1992, MacDougal described P. xiikzodz from herbarium specimens circulated as P. coriacea from Belize, Guatemala, and the Yucatán Peninsula. He found the fl oral corona of this new species to be fundamentally diff erent from P. coriacea and the other members of supersection Cieca, as it is 5-7-seriate as opposed to 2-seriate. He noted the absence of the fl oral nectary and the very reduced, denticulate operculum of this species. Th e seeds are also longer than all of the other species in the supersection. Th e petiolar nectaries are positioned on the distal third of the petiole in P. xiikzodz and the fl oral stipe is diagnostically long. MacDougal further separated P. xiikzodz into two subspecies, P. xiikzodz subsp. xiikzodz and P. xiikzodz subsp. itzensis. Th ough he found numerous diff erences in the fl owers of the two subspecies and artifi cial cross-pollinations between them proved unsuccessful, he felt that more information was needed to support the recognition of two separate species. I recognize the two species, P. xiikzodz and P. itzensis, which is supported by my morphological and molecular analyses of the taxa (see chapters 4 and 6).
Passifl ora itzensis and P. xiikzodz are identical vegetatively, but the fl owers are quite diff erent. Th e fl owers of P. itzensis lack or have a greatly reduced dark reddish purple androgynophore, are smaller, possess fewer fi laments in the outer and second coronal rows, an androecium and gynoecium with reddish purple pigmentation, very short styles, stigmas with their receptive surfaces presented distally, and anthers that do not fl ip over to an extrorse position after the fl ower buds open but move only slightly from the original introrse position to present their pollen distally. In the herbarium, it is not necessary to have perfectly preserved fl owers to diff erentiate between P. itzensis and P. xiikzodz, as the fl oral stipe of P. itzensis is commonly shorter than that of P. xiikzodz. Incidentally, in the dried fl owers of both P. itzensis and P. xiikzodz, the coronal fi laments appear nearly black.