A revision of Passiflora sections Insignes and ×Inkea ( Passifloraceae ) by

Jørgensen, P.M. & Vásquez, R. 2009. A revision of Passiflora sections Insignes and ×Inkea (Passifloraceae). Anales Jard. Bot.


Introduction
Our work on Passifloraceae in Bolivia (Jørgensen & al., in prep.) has led to the discovery of a new species that we call Passiflora carrascoensis P. Jørg.& R. Vásquez.The species clearly belongs to Passiflora subgenus Passiflora supersection Tacsonia section Insignes Feuillet & J.M. MacDougal, a small group of closely related species, best known in the group are P. pinnatistipula Cav. and P. mandonii (Mast.)Killip.P. pilosicorona Sacco also belongs in this section, although originally placed in supersect.Distephana (Juss.ex DC.) Feuillet & J.M. MacDougal.The sect.Insignes is almost exclusively Bolivian, with four species endemic to that country, one endemic to Peru, and the last, P. pinnatistipula, frequently cultivated from Colombia to Chile, with the highest concentration of collections from southern Peru and northern Bolivia.Because of the knowledge accumulated in conjunction with these discoveries we found it appropriate to convey what we have learned about the Passiflora sect.Insignes Feuillet & J.M. MacDougal and the related ×Inkea Feuillet & J.M. MacDougal.
The sect.Insignes was known for 125 years linked to the generic name Poggendorffia H. Karst.Two years after the publication of the genus, Karsten (1859, http://www.botanicus.org/page/825076)included a beautiful colorful illustration of Poggendorffia rosea H. Karst.Karsten's genus Poggendorffia and species Poggendorffia rosea have been treated in various ways and several times, and is central to understanding the history of the groups.
Six years later Bentham & Hooker (1862) reduced Karsten's genus to a synonym of the genus Tacsonia Juss.and they called Poggendorffia rosea a monstrous form of Tacsonia pinnatistipula (Cav.)Juss.Karsten (1887; see also Killip, 1938) reacted strongly to his genus being reduced to synonymy and his species called monstrous, but today it seems strange that he waited 25 years to publish his reaction.Was the death of Bentham in 1884, a coincidence or an explanation for his delayed reply?
In the mean time, Masters (1872) simply placed Poggendorffia rosea in synonymy of Tacsonia pinnatistipula, but that was probably an over simplification.Studying Karsten's illustration it is obvious that several key characteristics do not match Passiflora pinnatistipula.First, the stipules are not completely pinnatisect (if dissected at all), only the apical part of the stipule is dissected.Second, the bracts are much larger and may be fused at the base.Third, and most remarkably the stamens, with their odd appendages, have an opening on the adaxial surface at about the middle of the androgynophore and from that point upwards the stamens are free.Fourth, the basifixed anthers with a large apical connective are all characters not seen elsewhere in the entire genus Passiflora.The very different (referred to as monstrous by many) androecium was Karsten's (1856Karsten's ( , 1859) ) main argument for making a new genus.
Triana & Planchon (1873) did not accept the genus Tacsonia and changed the status of the genus Poggendorffia to a section of Passiflora within the subg.Tacsonia.They further stated that Karsten's genus was based on a monstrous specimen with some characters that belongs to sect.Bracteogama (now Elkea), and they also synonymized Poggendorffia rosea with Passiflora pinnatistipula.
Triana was an eyewitness to Karsten's discovery, and they (Triana & Planchon, 1873) wrote [translated from French] "One of us was with Mr. Karsten when, in a garden in Bogota, a Passifloreae caught our eyes" … "its five stamens had their filaments free since the middle of the column or gynophore; these filaments were concave like a dimple at their base and the anthers were deformed.This character of the androecium, described as normal by Mr. Karsten who could not have seen it but just in passing, was in fact nothing but accidental and monstrous.Indeed I observed that plant to collect seeds; at first the fruits from flowers with abnormal anthers aborted, then others became fertile because their androecium was normally built, that is: with filaments adherent to the column nearly to the ovary and with normally shaped anthers."We believe that Triana's observations come from two plants that may have been intertwined, one a typical P. pinnatistipula with normal stamens and anthers and the other a hybrid with abnormal stamens and anthers.They do not describe the shape or pubescence of the fertile or aborted fruits.Karsten (1887) Sodiro (1903) made the combination Tacsonia ×rosea (H.Karst.)Sodiro and accepted the "species", but was convinced that it was a hybrid between P. pinnatistipula and P. mollissima (Kunth) Bailey (today P. tripartita var.mollissima (Kunth) Holm-Niels.& P. Jørg.).Both species were, then as now, frequently cultivated and often close together.He used a multiplication sign in front of the "species" name and explained that he had numerous specimens that agreed with Karsten's description of the androecium.Harms (1925) accepted seven unranked groups within Passiflora sect.Tacsonia (Juss.)Triana & Planch.(supersect.Tacsonia today).Two are of interest in this context.The fourth group was named Pinnatistipulae Harms including P. pinnatistipula.He also included Poggendorffia rosea here and cited Sodiro's observation of it being a hybrid with P. mollissima.His fifth group was called Insignes Harms and was based on P. insignes (Mast.)Hook.f.The difference between Harm's two groups was entire versus three-lobed leaves.Killip (1938) also agreed with Sodiro, maybe due to his own field observation of a plant of Passiflora ×rosea (H.Karst.)Killip at Tarma in Peru growing between a plant of P. pinnatistipula and P. tripartita var.mollissima.Killip did not accept any subdivisions of the subgenus Tacsonia and listed Poggendorffia as a synonym of Tacsonia.Escobar (1980) used the name sect.Poggendorffia (H.Karst.)Triana & Planch.for a grouping of species that included Passiflora insignes, P. mandonii, P. pinnatistipula, and P. ×rosea.In 1988, she added P. pilosicorona to this section.The section has several unique characters and can be postulated as a monophyletic group within the supersect.Tacsonia if the hybrid is excluded.
In summary, Bentham, Hooker, Masters, and Triana and Planchon, would apparently have liked to "forget" about Karsten's plant by simply calling it a monstrous individual and relegating it to synonymy.Sodiro, Harms, Killip, Escobar, and Feuillet and Mac-Dougal were in our opinion correct in accepting it as a hybrid.Strangely enough nobody has, to our knowledge, tried to cross the two parents to confirm that the hybrid's F1 morphology of the stamens is as indicated by Karsten. Feuillet & MacDougal (1997) pointed out that if Passiflora ×rosea is a hybrid, all names based on this type are consequently hybrids, i.e. the genus and sect.Poggendorffia.The type contains genetic material from two species belonging to two different sections (now named Insignes and Elkea) and logically it cannot be restricted for use to only one of the groups, as done by Escobar (1980Escobar ( , 1988)).Section Poggendorffia can only be the logical name if both parents are included (and then it would not be the oldest name).This, most logical, but unusual situation, simultaneously creates the dilemma that Passiflora ×rosea cannot be accepted within a section in a treatment that recognizes sect.Insignes or Elkea.The name Poggendorffia does not conform to the International Code of Botanical Nomenclature article H. 7 (McNeill & al. 2006) and in consequence Feuillet & MacDougal established the hybrid sect.×Inkea with its only member Passiflora ×rosea (H.Karst.)Killip.They did not place Poggendorffia in synonymy of sect.×Inkea, which we interpret as an omission.

Materials and methods
A total of 233 collections have been revised from AAU, BM, BOLV, CUZ, G, Herbarium Vásquez, K, LPB, MA, MO, P, QCA, S, US, USM, and USZ.Herbarium Vásquez is a private collection housed in the residency of the junior author.We have occasionally cited specimens that we have not seen to document the entire distribution of the species as we collectively know it at present.The non revised specimens are clearly marked and only included when they A revision of Passiflora add a political unit to the specimen citation; and this information comes from either Escobar (1980) or Killip (1938).An appendix list all numbered collections with the species they belong to.Our emended descriptions are short and only include the most important and characteristic features.They are largely based on the literature, for the species Passiflora insignis, P. lanceolata, P. mandonii, and P. ×rosea we particularly consulted Escobar (1980), and for P. lanceolata and P. ×rosea we also included observations by Killip (1938).We do not know if fig. 7 was documented with a voucher and the photographer has not responded to requests for information.Lianas, often densely pubescent; leaves entire or three lobed; stipules deeply pinnatisect or divided into filiform segments.Bracts free to base with serrate irregular margins; floral tube of varying length; sepals carinate; petals slightly shorter than sepals; corona in one or two series, the outer series well developed or occasionally reduced to a tuberculate ring, acicular 1-2 cm long, the inner series, reduced to few denticulate projections, placed in the floral tube, between the operculum and the mouth, missing in some species or collections, but a slight difference in coloration often indicating where the lost corona originally was placed.Fruits are spherical or almost so in all species with known fruits.
Morphology and affinities.Killip (1938) placed a very high taxonomic value on the presence and fusion of a second corona row.This character was used, not very convincingly, to separate supersect.Distephana (Juss.ex DC.) Feuillet & J.M. MacDougal and sect.Manicata (Harms) Feuillet & J.M. MacDougal from sect.Tacsoniopsis Triana & Planch., Rathea (H.Karst.)Harms, and Tacsonia (Juss.)Harms (the last four names currently placed in supersect.Tacsonia, but were in Killip's treatment all considered subgenera).Killip (1938) placed, for instance Passiflora callimorpha Harms in subg.Distephana while P. insignis was placed in subg.Tacsonia the two names are currently seen as synonyms of the same species with some variation in the formation and loss of a second corona series.There are obviously some similarities between supersect.Distephana and Tacsonia, in particular sect.Insignes.Furthermore, the presence of a well developed corona is in Tacsonia seen as a primitive character, and it is tempting to present a hypothesis that suggest that Tacsonia have originated from Distephana and as the group adapted to the montane environment it spread northward and became increasingly diversified in Peru, Ecuador, and particularly Colombia.The species we have left today and may support such a hypothesis are Passiflora buchtienii Killip and P. miniata Vanderplank.They are members of supersect.Distephana, but found at relatively high elevations in the Bolivian Andes.Representing an unusual distribution in the supersection that is elsewhere exclusively lowland.They may have shared an ancestor with sect.Insignes.Further adaptation took place while moving northward resulted in losing the well developed corona.Parallel to the morphological changes a strong diversification took place within supersect.Tacsonia that now includes about 45 additional species.
Diagnostic characters.Escobar (1980) discussed in detail the minor differences found between the type of Passiflora callimorpha and typical P. insignis showing that variation is found in several features that Harms used to segregate the species, notably the one or two corona series and the fusion of the inner series into a tubular membrane; see also the discussion of morphology and affinities of the section.The species is easily recognized within the section by its entire leaves with densely pubescent undersurface.
Diagnostic characters.Passiflora mandonii has the longest floral tube of all the species in the section, shorter sepals and petals that are not reflexed, but form a bell shaped structure surrounding the reproductive organs.It shares a three lobed leaf with triangular lobes with P. carrascoensis, but apart from that species the leaf shape is very characteristic and can not be mistaken for any other species.
Distribution and habitat: Passiflora mandonii is known from a series of localities in the Yungas of Bolivia from Cochabamba to La Paz.One collection comes from Potosí, possibly cultivated, and one from Puno in Peru that is also from a cultivated plant; further north a single specimen from Cuzco of a white form has been apparently naturalized in roadsides.We know of white forms of P. pinnatistipula and this could potentially be a hydrid.An IUCN threat status of NT is here assigned to this species.
Diagnostic characters.Passiflora pinnatistipula shares a 3-lobed leaf with P. mandonii, P. carrascoensis, and P. pilosicorona in this section.It shares a longer than 5 cm hypanthium only with P. mandonii, and is distinct from that species by the color of the indument, the length of the hypanthium, and the shape of the leaf lobes.
Distribution and habitat.Passiflora pinnatistipula is widely distributed, from Chile and Bolivia to Colombia, in large part due to its frequent cultivation.Our best guess for its natural distribution would include areas in northern Bolivia and southern Peru, where it is most frequently collected.An IUCN threat status of LC is assigned to this species.
Liana climbing 4-6 m over trees and shrubs, pubescent except for the adaxial surface of leaves, internal floral parts and petals; indument light brown, tangled.
Diagnostic characters.Escobar (1980) saw two specimens with short floral tube and shorter corona filaments, but she discussed them with Passiflora mandonii and did not indicate collector and number of the deviating specimens.She remarked that there were "no gradual intergradation between the typical P. mandonii and these specimens", and saw them as possible hybrids with P. insignis.Those specimens may well belong to the new species here described; if that is the case then it is strange that she did not note that the sepals and petals are about twice as large as in P. mandonii.This new species is, if it is of hybrid origin, more likely to be between P. mandonii and P. pilosicorona.
Passiflora carrascoensis and P. pilosicorona have very similar leaves, color of the indument, and a outer corona that is curved towards the androgynophore, but differs through the shape of the bracts (ovatelanceolate versus broadly ovate), and the floral tube length that is normally longer.It differs from P. mandonii by the shape of the leaves (shorter in P. carrascoensis), by the length of the sepals and petals (7 and 6 versus 6 and 5 respectively), by the bracts (ovatelanceolate versus ovate), by the stipules (more branched in P. mandonii), by the outer corona bent towards the androgynophore, and by the peduncles 4-7 cm versus 6-15 cm.
Distribution and habitat.Passiflora carrascoensis has a very restricted distribution in a rather small area along the road from Cochabamba to Villa Tunari, at 98-125 km from Cochabamba and from the region of Sehuencas (near the type locality of P. pilosicorona), another 130 km from Cochabamba towards Santa Cruz along the old road.The species has only been collected 12 times in the last 45 years and to our knowledge, not before then.Two of the cited collections from Riksherbariet in Stockholm (S) could have been seen by Escobar, but they were never annotated by her.An IUCN threat status of EN is here assigned.6-12(17) cm long; bracts (3)3.5-4× (2)2.5-3cm, prominent, membranaceous, broadly ovate to almost orbicular; apex acute; base cordate; margins irregularly serrate and glandular; stipe to 1 cm long; floral tube 2.0-2.2(4)× 1 cm; nectar chamber dilated at base, 8 mm long, 20 mm wide, ferrugineous lanate externally; sepals and petals spreading, dark rose-colored; sepals 5.5-6 × 1.2 cm, oblong-lanceolate, membranaceous, dorsally carinate, adaxally villosous, aristate; awn 7-10 mm long; petals 5.5 × 1.3 cm, oblong, apex rounded; corona in 3 series; the outer series falcate, incurved, to 10 mm long, blue, the ones born at the base of the sepals dorsally pilose, the ones from the base of the petals glabrous; the middle series composed of tuberculate process, 1 mm long, white, located near the base of the floral tube; the inner series compose of an erect, tubular, filamentose membrane, 5 mm long; operculum tubular, reflexed membrane, with serrulate margin; androgynophore 4 cm long; ovary obovoid, 10 × 6 mm, densely pubescent, indument white, the base stipitate.Fruit subglobose, 6 × 6.5 cm.Seeds pitted, 7-8 mm long, 4.5 mm wide, the apex mucronate.(Fig. 6).
Diagnostic characters.Passiflora pilosicorona is recognized by the large bracts, the short floral tube, the deep rose color of the petals, the outer filaments of the corona with short pubescent processes (as the name pilosicorona implies) and the elongated peduncles in the fruiting stage.
Distribution and habitat.Passiflora pilosicorona is known to grow in two separate geographical areas in Bolivia.The first is along the border between the departments of Cochabamba and Santa Cruz on the old road.The vegetation is an evergreen humid montane forest or scrubland.The area is also known as Siberia because of its cold climate with clouds and fog that almost always cover it.The other area is located some 15 km E of the city of Vallegrande with almost similar climatic conditions and forest formation.An IUCN threat status of EN is assigned to this species.separation from the androgynophore; androgynophore 1-3 cm, thin; gynophore 3-4 cm, stout; ovary ovoid; stamens partially free, the stamens separation area is shaped in several different ways (see photography), anthers basifixed or almost so, pollen scarce; connective awn shaped.Mature fruits 8 × 2.5 cm, ovoid oblong; seeds 6 mm, closely reticulate.(Fig. 7), see also http://www.botanicus.org/page/825076.Diagnostic characters.A vigorous liana with well developed corona as in P. pinnatistipula while the leaf shape and pubescence is more similar to P. tripartita var.mollisima.The stipules seem to either take after one species or the other.The most characteristic feature of this hybrid is the unique and "monstrous" androgynophore and base of the stamens which seem to take different shapes as well; we are not at all sure how to interpret the structure except as an indication that the two parents may not be as closely related as we think.Hybrids are readily produced in many areas of Subg.Passiflora and without such remarkable results.

Additional specimens examined
Distribution.P. ×rosea can be found from Colombia to Chile and Bolivia, always under cultivation or near settlements frequently seen in close proximity of one or both its parent species.The native range of the parent species would probably not overlap, but that is difficult to ascertain as both species have been widely cultivated for a long time.P. pinnatistipula is probably native to southern Peru and northern Bolivia (maybe also northern Chile) while P. tripartita var.mollisima is a cultivar mainly from Colombia to central Peru.IUCN threat assignment is LC, parent species are frequently cultivated.
Note on typification.The material at P, which Killip annotated in 1927 as type material and cited (1938, p. 279), is not type material.The label clearly indicates that it was collected by J. Triana, in the "environ de Bogotá", i.e. not a collection made by Karsten.This material was not seen by Escobar (1980), she however saw material from W, and indicated that as the holotype.Karsten worked, however, in Berlin in 1856 (TL-2, http://tl2.idcpublishers.info/consulted 12-VIII-2008) so it would be most logical that the holotype would have been deposited there, as indicated by Killip (1938), but that specimen was destroyed.Escobar (1980) is furthermore not effectively published and her indication of a holotype can therefore not stand as a lectotypification, so the lectotypification is therefore made here.

Fig. 4 .
Fig. 4. Passiflora carrascoensis: a, leaf and stem, with detail of pubescence on upper surface; b, lateral leaf lobe tip with detailed pubescence on lower surface; c, stipules; d, bract; e, petal; f, sepal; g, longitudinal section of flower; h, longitudinal section of base of floral tube [a-g from S. Arrázola 166].

Fig. 7 .
Fig. 7. Passiflora × rosea: A, gynoecium and androecium, note the deform anthers with connective, the trochlea like development on the androgynophore where the filaments separate from the gynophore; B, longitudinal section of flower and bud; C, axil with pedicel and flower bud with fused bracts; D, immature fruit, note the ovoid fruit and the trochlea like structure still present in fruit.Scale bars: a-d = 1 cm.
.M. Jørgensen & R. Vásquez Distribution and habitat.Only known from the type from what Escobar interpreted as from Andamarca in Junín, Peru.That interpretation is probably correct, the label states "Mount Allis Andimarca July" and Mathews 1230 (the type of Ancylogyne capitata Nees) is from Junín P . 6: 393.1805.Passiflora pennipes Sm. in Rees Cycl.26: Passiflora no.