Remarkable disjunctions in Ipomoea species (Convolvulaceae) from NE Brazil and Central America and their taxonomic implications

Summary Recent collections of Ipomoea from North East Brazil have revealed a number of unexpected disjunct distributions. The most remarkable is that of I. eremnobrocha D. F. Austin, previously thought to be endemic to Panama but now known from three states in NE Brazil. Revision of Panamanian material named I. eremnobrocha unexpectedly showed that two distinct species had been treated under this one name. Specimens from the Chagres National Park area in Panama are described as a new species under the name I. isthmica J. R. I. Wood & Buril while I. eremnobrocha is retained as the correct name for the plant from the Altos de Campana in Panama and NE Brazil. An amended description of this species is given and a table of differences between the related species is provided. Two recently described species from Bolivia, I. graniticola J. R. I. Wood & Scotland and I. chiquitensis J. R. I. Wood & Scotland are recorded from NE Brazil several thousand km from their type localities. Attention is drawn to the role of granite inselbergs as sites of species with a disjunct distribution. A possible relative of I. chiquitensis is described as a new species from NE Brazil under the name I. melancholica J. R. I. Wood & Buril. The new species are illustrated with line drawings and maps of the unusual distribution patterns are provided.


Introduction
Northeastern Brazil is drier than other parts of the country and much of it is covered in various kinds of dry forest and, in particular, with deciduous, often spiny woodland known as caatinga. There are exceptions to this picture. South from approximately 7°S and running parallel to the coast are relics of Mata Atlântica or Brazilian Atlantic Forest (Nogueira Rodal et al. 2005), which in some areas penetrates into the more widespread caatinga of the region. The Mata Atlantica itself is not uniform but consists of a mosaic of different physiognomies including deciduous woodland and, at higher elevations with greater rainfall, a distinctive vegetation known as "Brejo de Altitude", sometimes translated as "northeastern upland forest" (Porto et al. 2004). As in other parts of Brazil diverse species of Ipomoea L. are present and it was from here that the authors found two putative new species which are discussed in this paper and whose distribution and identity has extensive ramifications.

Materials and Methods
Our results are based on the examination of dried material principally from herbaria in Brazil and the United States.
We have been able to compare specimens with a wide range of material from elsewhere as part of on-going monographic work on Ipomoea being carried out at the University of Oxford. We have also made use of on-line images, principally from Jstor http://plants.jstor.org. and from the herbarium of the Universidad de Panama http:// herbario.up.ac.pa/Herbario/herb/vasculares/ view.species/1392. The Tropicos data base http:// www.tropicos.org/ has also provided useful information. Field work in Brazil was carried out by Teresa Buril and students. Molecular sequencing of samples was carried out at the Department of Plant Sciences at Oxford.

The identity of Ipomoea eremnobrocha
One of the putative new species was quite unlike any other species recorded from Brazil. It had deeply 3lobed leaves which were sericeous on the lower surface, a shortly pedunculate, compact cymose inflorescence with a cream-coloured campanulate corolla and densely woolly seeds. This suggested an affinity with the Mesoamerican species Ipomoea peteri (Kuntze) Staples & Govaerts but this could be easily distinguished by its larger, pink, funnel-shaped corolla and by the much longer outer and inner sepals. We were confident that we had a very distinct new species.
We had not carefully compared the putative new species with Ipomoea eremnobrocha D. F. Austin, described from Panama, because the entire-leaved plant illustrated with the protologue (Austin 1997: 152) was so obviously different from the plant in NE Brazil. It was only when the first author obtained a loan from Missouri (MO) as part of Oxford University's monographic studies of Ipomoea, that it was realised that specimens from Panama identical to the Brazilian plant had been named I. eremnobrocha by Austin. A further loan made it clear that Austin had labelled specimens of two separate species from Panama with this name and that the protologue of I. eremnobrocha was based on elements of both species.
It was accordingly necessary to see the type of Ipomoea eremnobrocha (A. Gentry 5759) in order to decide which species the name was linked with. Here, however, another complication arose as the holotype could not be found at MO, nor the isotype at ARIZ or FTG, to which Austin's Convolvulaceae specimens had been transferred from FAU. No duplicate was present at PMA either. It must be assumed that the original material was never returned to Missouri and has somehow been mislaid during the successive transfers of Austin's Convolvulaceae from FAU to FTG and thence to ARIZ.
RECOGNITION. Ipomoea isthmica is a distinctive species not readily confused with any other species from Mesoamerica because of its liana habit, large, abaxially sericeous, entire leaves and shortly pedunculate compact inflorescence with a funnel-shaped magenta corolla and woolly seeds. It might be compared with I. steerei (Standl.) L. O. Williams as both species have entire leaves but the relatively small (<6 × 3.5 cm) oblong-elliptic, strongly apiculate leaves of I. steerei are very different from the ovate to suborbicular, acute, retuse to shortly acuminate leaves of I. isthmica which are 7 -15 × 6 -11 cm in size. Additionally, the lax fewflowered inflorescence of I. steerei is quite different. I. isthmica is distinguished from I. eremnobrocha by the much larger, entire, not 3-lobed leaves, the longer very unequal sepals and the longer pale magenta corolla with a dark purple base. These differences can be observed in the comparison plate, Fig. 1, and in Table 1. Vegetatively I. isthmica appears most similar to I. cuprinacoma E. Carranza & J. A. McDonald in the size and shape of the leaves and in the distinct shape of the sepals. However I. isthmica differs in the compact, up to 10-flowered cymes with short peduncles and pedicels, relatively persistent bracteoles, the suborbicular inner sepals and the magenta corolla with blackish throat. In contrast I. cuprinacoma has relatively lax 1 -3 (-5)-flowered cymes with pedicels up to 25 mm long, early caducous bracteoles, obovate inner sepals and a slightly larger (to 8 cm long) white corolla with bluish throat. ILLUSTRATIONS. There is a good illustration of Ipomoea isthmica accompanying the original protologue of I. eremnobrocha (Austin 1997: 152). All the elements in this plate and the specimens from which they were drawn are clearly attributable to I. isthmica. The description, in most cases, of the corolla as pink with a dark centre, rather than white confirms this. Folsom's description on the collection label as "Corolla antique pink turning green to a black throat" is particularly  HABITAT. The habitat of Ipomoea isthmica is variously described as "premontane wet forest" or "cloud forest" and it was collected on forest edges and in roadside scrub between 300 and 1000 m. CONSERVATION STATUS. Although geographically restricted, all or most of the c. 20 records of this species are from the Chagres National Park, which enjoys legal protection. The new species favours disturbed habitats so should face no significant threat. However in the complete absence of any population studies, this species can only be categorised as Data Deficient (DD) until such studies can be carried out. NOTES. We have not yet been able to sequence material of this species but its morphology suggests it belongs to the small clade with Ipomoea steerei, which is discussed below. As the protologue of I. eremnobrocha (Austin 1997: 145ff.) combines elements of I. isthmica with I. eremnobrocha, we are redescribing the former with an amended description as follows: Ipomoea eremnobrocha D. F. Austin (1997: 145), emend. J. R. I. Wood & Buril. Type: Panama, Cerro Campana, A. Gentry 5759 (holotype cited from MO and isotype cited from FAU, but both missing).
Perennial climber or liana of unknown height but reaching at least several metres high, stems twining, somewhat woody below, herbaceous above, thinly pubescent when young, glabrescent, pale brown. Leaves petiolate, ovatetruncate in outline, 5 -12 × 7 -12 cm, 3-lobed to about half way, base ± truncate or subcordate and shortly cuneate onto the petiole, apex finely acuminate and shortly mucronate, central lobe oblong-elliptic (rarely ovate), 2 -5 × 2 -4 cm, laterals broadly ovate, margins entire or undulate, adaxially green, pubescent, abaxially densely silvery-sericeous with appressed hairs and scattered white glands; petioles terete, 4 -11 cm, thinly pubescent. Inflorescence of compact axillary cymes with up to about 10 flowers; primary peduncles 0.5 -2.5 cm, stout, pubescent; bracteoles 2 -7 × 0.5 -1 mm, filiform to lanceolate, acuminate, pubescent, tardily deciduous; secondary peduncles 3 -5 mm; pedicels 3 -8 mm, pubescent; sepals somewhat unequal, outer 4 -5 × 2 -3 mm, broadly oblong, truncate or slightly retuse, glabrous or with a few hairs at the base, inner 5 -6 × 3 -4 mm, obovate, usually strongly retuse with a broad sinus so appearing winged, margins scarious; corolla ±campanulate, white, glabrous on the exterior, 2 -2.5 cm long; limb 2.2 -2. 5 cm diam., stamens included; filaments inserted just above the corolla base, glabrous except for pubescent base, longer 9 mm, shorter 8 mm, anthers 2 mm long; style 15 -18 mm, glabrous, stigma biglobose; ovary glabrous. Capsule 12 -13 × 10 -11 mm, ellipsoid to subglobose, very shortly apiculate with persistent style base, glabrous, 4-seeded; seeds 6 × 1.5 -2 mm, brown, broadly oblong in outline, densely lanate with matted cottony hairs up to 10 mm long. Fig. 1H -N. RECOGNITION. Ipomoea eremnobrocha is most closely related to I. peteri, which is widely known under the name I. tuxtlensis House. Austin (1997) correctly compared it with this species as both have 3-lobed leaves, abaxially white and dotted with white glands combined with compact, shortly pedunculate inflorescences with somewhat persistent lanceolate-filiform bracteoles and woolly seeds. I. peteri can be distinguished easily by its larger (c. 5 cm long), funnelshaped pink corolla and the subequal, densely hirsute sepals. Austin also compared I. eremnobrocha with I. steerei but this species is immediately distinguished by its oblong-elliptic, entire leaves, lax, few-flowered inflorescence and relatively large, funnel-shaped pink corollas c. 5 cm in length. I. isthmica described above is also readily distinguished by the large, ovate to suborbicular, entire leaves, longer sepals and the larger funnel-shaped pale magenta corolla with a dark tube. The 3-lobed leaves with finely attenuateacuminate tips, short, basally pubescent sepals and the smaller subcampanulate, cream-coloured corolla of I. eremnobrocha are very distinct.  It is unfortunate that the epithet "eremnobrocha" has to be used for this species as it is based on the Greek "eremnos" = black and "brochtos" = throat (Austin 1997: 148) which is very appropriate for Ipomoea isthmica but quite inappropriate for this species. At Oxford we have been able to sequence material of Ipomoea eremnobrocha using ITS from M. D. Correa et al. 11312 and this shows that it belongs to a wellsupported clade with I. peteri, I. steerei and I. heterodoxa Standl. & Steyerm., all four species from Mesoamerica. Given its morphology, it is likely that I. isthmica also belongs to this clade. The differences between these four species are shown in Table 1.

A note on disjunct distributions
The most striking disjunctions in neotropical Ipomoea are those of a number of relatively common species with a distinct amphitropical distribution, something only occasionally commented on in the literature, for example McDonald 1995: 112. These species are centred in the northern hemisphere on Mexico and the United States and in the southern hemisphere on Bolivia, Argentina and Paraguay. They are absent from all or most of the intervening territory. Among species showing this kind of distribution are I. amnicola Morong, I. cordatotriloba Dennst., I. crinicalyx S. Moore, I. plummerae A. Gray and I. pubescens Lam. Although there is some suspicion that the first two were introduced over part of their range, there is no dispute over the status of the other three.
A unique disjunct distribution is that of Ipomoea subrevoluta Choisy. It is entirely restricted to the South American continent apart from single occurrences on the island of Trinidad and Isla de Juventud (or Pinos, as it was formerly known) in Cuba. As it is a species of semi-aquatic habitats it might be an ancient introduction to Cuba by birds. No such explanation seems possible for the disjunct distribution of I. eremnobrocha. The limited molecular data suggest this species is of Mesoamerican origin and its presence in NE Brazil cannot be explained at the present time by any of the scenarios suggested, for example, by Gentry (1982). We are unaware of any similar distribution pattern except that of Jacquemontia gracillima (Choisy) Hallier f. which occurs in scattered populations in NE Brazil (and also in eastern Bolivia) but is only known in Central America from a single collection (H. Pittier 4842 (F)) made close to but not in the Altos de Campana National Park in Panama.
Two other species of Ipomoea have been discovered in NE Brazil recently, both described from Bolivia in 2015 (Wood et al. 2015). I. graniticola known from two granite domes in Bolivia and a quartzite outcrop in Mato Grosso (Wood et al. 2015) has recently been found in the State of Ceará (E. B. Souza et al. 3395 (HUVA, PEUFR,) in a very similar habitat. I. chiquitensis, which was known from a single granite rock platform near San Rafael in Eastern Bolivia, has also been found in Ceará (J. A. A. M. Lourenço 124 (PEUFR)). Although, the disjunction in these two cases is of almost 3000 km, it is not quite so surprising. The rock domes and platforms which are the habitat of these species function as inselbergs and harbour assemblies of plants very different from the surrounding vegetation but similar in composition to other inselbergs several hundred km away. The remarkable feature in this case is the distance between the known Map 2. Distribution of Ipomoea melancholica (♦) in NE Brazil.
populations. In passing it is worth noting that another species of Ipomoea, I. caloneura Meisn., occurs on isolated inselbergs in Eastern Bolivia and neighbouring parts of Mato Grosso. It has not yet been found in NE Brazil.
RECOGNITION. This species has been interpreted as a form of Ipomoea acanthocarpa (Choisy) Aschers & Schweinf. (=I. piurensis O'Donell) but differs in the solitary flowers and very shortly rostrate capsule. It has also been identified as I. minutiflora (M. Martens & Galeotti) House but differs in its larger solitary pink flowers and larger capsule, 10 mm, not 3 mm in length. It might also be thought to be a depauperate species from the Pharbitis group such as I. indica (Burm.) Merr. but the 4-seeded capsule and small sepals rule that out. The species seems closest to I. chiquitensis, a similarly modest, usually solitaryflowered species with a small corolla. Both species have leaves adaxially pubescent but abaxially glabrous, and both have similar-sized, acuminate sepals with white margins as well as deflexed fruiting peduncles. However, I. chiquitensis always has entire leaves, the stem is glabrous but the leaves and sepals are much more hirsute, and the capsule is much more prominently rostrate. HABITAT & DISTRIBUTION. Recorded from two states and three localities in northeastern Brazil (Map 2). very little is known about its habitat but the type appears to have been collected around a rock outcrop in Atlantic Forest, a similar habitat to that of Ipomoea chiquitensis. It grows in dry forest fide Araujo 1424. With only three relatively recent collections this species appears to be rare despite its presence in two states. However, it can only be classified as Data Deficient (DD) within IUCN (2012) guidelines until its populations can be studied in detail. ETYMOLOGY. The epithet melancholica ("sad") refers to the drab appearance of this species with its solitary flowers, usually bent-over in fruit, thus presenting a rather forlorn aspect.