Generic status of Quisqualis (Combretaceae), with notes on the tax- onomy and distribution of Q. parviflora

The taxonomic status of Quisqualis L., a genus closely related to and sometimes considered congeneric with Combretum Loefl ., is discussed. The genus Quisqualis is maintained to accommodate selected African, Indian, Asian and Malesian species (including Q. indica L. and Q. parvifl ora Sond.). Diagnostic characters for Quisqualis include: leaves with abundant subepidermal crystalliferous idioblasts, each containing a large druse of calcium oxalate; leaves, stems and fl owers with stalked glands but no scales; persistent petiole bases that develop into curved spines; petals well developed; hypanthium tubular or cylindrical, usually longer than 20 mm; stamens and style not or scarcely exserted; style adnate to hypanthium for more than half its length. Quisqualis shares a number of morphological similarities with Combretum Loefl . subgen. Cacoucia (Aubl.) Exell & Stace sect. Poivrea (Comm. ex DC.) G.Don. Some species of Combretaceae from West, West Central and East Africa have the style adnate to the upper hypanthium and display features reminiscent of both Quisqualis and Combretum. These species also have characters of their own and in the past were placed in different sections under Combretum. It is suggested that at least some of these species may be best classifi ed in genera distinct from Combretum and Quisqualis, one of which is Campylogyne Hemsley. Combretum s.str. is defi ned on the basis of a combination of characters and includes species of which the upper hypanthium is variable in shape, but when tubular or cylindrical, then always shorter than 20 mm. Other diagnostic characters include: stamens exserted well beyond petals; style exserted and free, but when shortly adnate to upper hypanthium (only at the base or for a short distance), then stamens long-exserted. It is suggested that different pollination strategies have developed independently in the Combretum–Quisqualis clade, resulting in convergent morphological trends in fl oral morphology. These homoplasious similarities in fl oral morphology are at the root of the diffi culties experienced to demarcate genera. An alternative classifi cation is provided for those preferring to include the southern African Quisqualis parvifl ora under Combretum s. l. For this purpose, a new combination and name, Combretum sylvicola O.Maurin is provided. Quisqualis parvifl ora is confi ned to the Eastern Cape and KwaZulu-Natal coastal regions and does not extend beyond this area as has been claimed by some. A comparative table to differentiate among four groups in Quisqualis and Combretum in Africa, as well as a photo of a herbarium specimen and a distribution map of Quisqualis parvifl ora, are provided.


INTRODUCTION
Quisqualis L ., a genus comprising ± 17 species, occurs in Africa, India, Asia (China) and Malesia (Exell 1954;Chen & Turland 2007) .It shares a number of morphological similarities with Combretum Loefl .subgen .Cacoucia (Aubl .)Exell & Stace sect .Poivrea (Comm .ex DC .)G .Don .In southern Africa, Combretum mossambicense (Klotzsch) Engl .and C. bracteosum (Hochst .)Brandis belong to this section .Quisqualis parviflora Sond .(Figure 1), the only southern African member of the genus, as well as the latter two species of Combretum, are characterized by the absence of epidermal glandular scales and leaves with stalked glands on the petiole, midrib, lateral veins and flowers .Scales are always present in subgen .Combretum where they are of considerable taxonomic significance, especially at sectional level (Jordaan et al. 2011) .Stalked glands, on the other hand, are of little use for resolving sections in subgen .Cacoucia as well as in genera such as Quisqualis and Calopyxis Tul .(Stace 1980) .Long unicellular combretaceous hairs (non-glandular, sharp-pointed, thick-walled with a bulbous base) are present in both subgenera of Combretum and in Quisqualis .Hennessy (1991) described the leaf lamina of Combretum bracteosum as 'pellucid-punctate' .This distinctive character has now also been observed in C. bracteosum, C. mossambicense, Quisqualis parviflora and Q. indica L .These so-called pellucid dots are abundant, spherical, subepidermal, crystalliferous idioblasts, each containing a large druse of calcium oxalate (Figure 2) (see Tilney 2002) .These idioblasts are also quite noticeable in dry material as numerous tiny bumps on one or both lamina surfaces when viewed under a stereo light microscope .Quisqualis may have extrafloral nectaries (Tilney & Van Wyk 2004), structures not yet recorded in Combretum .Furthermore, C. mossambicense, C. bracteosum and Quisqualis parviflora all have hooked or straight spines derived from persistent petioles by means of which they climb in or over vegetation .The flowers vary from white with long protruding stamens in C. mossambicense, bright red in C. bracteosum, to greenish with included stamens in Quisqualis parviflora.C. mossambicense has 5-winged, softly hairy fruits, but those of C. bracteosum are quite different in being wingless, hairless and indistinctly 5-angled nuts .
Quisqualis parviflora, according to Carr (1986), has 5-winged fruits .However, we could not find any fruiting material of this species in South African herbaria and it is possible that Carr inferred the fruit morphology from generic descriptions which are based mainly on non-African material .The 4-winged fruit attributed to Quisqualis parviflora and depicted in plate 1925 of The Bothalia 41,1 (2011) Flowering Plants of Africa (Carr 1986) is said to have come from Mariepskop, Mpumalanga .Considering the known distribution of this species in southern Africa (Figure 3), the fruit most probably belonged to a member of Combretum and not to Quisqualis .Tony Abbott (pers .comm .)has confirmed that he has not seen any fruits on this species in the Umtamvuna Nature Reserve for the last 30 years .As Carr (1986) states, even flowering seems to be a rare event, only a handful of flowering specimens exist in South African herbaria .On the other hand, the paucity of flowering material in herbaria may well be due to the fact that the flowers are inconspicuous and borne in the canopy of tall forest trees well out of the reach of plant collectors .The flowering and reproductive behaviour of Q. parviflora is clearly in need of further investigation .

History of the genus Quisqualis
The genus Quisqualis was established by Linnaeus (1762) .Hooker (1867), Lawson (1871) and Brandis (1898) separated Quisqualis from Combretum on the basis of its elongated, tubular, upper hypanthium which is subterete throughout and not constricted towards the base-a character state absent in typical Combretum .They attributed to Quisqualis several species from Asia, tropical and southern Africa .Engler & Diels (1900), on the other hand, recognized Quisqualis based on the alleged presence of dehiscent fruit, compared to indehiscent fruit in Combretum .However, this character proved to be unreliable because many species of Combretum have tardily dehiscent fruit .Exell (1931) proposed a new circumscription of Quisqualis and Combretum based on the insertion of the style in the upper hypanthium: if the style is adnate to  Exell's (1931) new definition of Quisqualis, some specimens from tropical Africa classified under Combretum were found in which the style is very shortly adnate to the upper hypanthium .To avoid this ambiguity, Exell & Stace (1964, 1966) redefined Quisqualis and separated it from Combretum by a combination of two characters: 1, adnation of the style to the hypanthium and 2, the non-exertion of stamens from the flower .Exell & Stace (1964) provided a useful key to distinguish between Quisqualis and Combretum .Jongkind (1991), on the other hand, proposed the amalgamation of Quisqualis with Combretum based on the adnation of the style to the upper hypanthium which he encountered in some otherwise undisputed species of Combretum .He formally transferred a number of species from Quisqualis to Combretum (Jongkind 1991(Jongkind , 1992(Jongkind , 1999) ) .Based on leaf anatomical evidence, Tilney (2002) supports the classification of Jongkind (1991) .Stace (2007) and Mabberley (2008) follow Jongkind's lumping of the two genera .

Quisqualis in Africa
The style is free from the hypanthium in nearly all species of Combretum in Africa and only in a few is there a certain degree of adnation to the upper hypanthium .For example, C. ghesquierei Liben from the Democratic Republic of Congo (DRC) has 4-merous (mostly 5-merous in other species) flowers with the style only adnate to the very base of the upper hypanthium, but with long-exserted stamens .Liben (1968) separated this species from all of the other members of Combre-tum in West Central Africa and placed it in a group of its own .Furthermore, Jongkind (1990) found that the style is fused to the upper hypanthium for more than 2 mm in C. grandiflorum G .Don from West tropical Africa (Guinea, Liberia and Sierra Leone, through Ivory Coast to Ghana) .It has bright red, 5-merous flowers with the upper hypanthium infundibuliform, 12-20(-25) mm long, glabrous or sparsely hairy, petals 10-15 mm long, glabrous or with few hairs on veins outside, but with the stamens long-exserted .Engler & Diels (1899) placed this species, together with other West African species, in a section of their own, namely sect .Grandiflorae, based on the nearly glabrous petals.Stace (1981) placed this section in synonymy with subgen .Cacoucia sect .Poivrea, a step which seems morphologically poorly supported .Although C. grandiflorum also has leaves with abundant crystalliferous idioblasts as in C. mossambicensis and C. bracteosum, and bright red flowers as in C. bracteosum, it has no stalked glands and almost glabrous petals .The petals in the latter two species are densely hairy .With its long-exserted stamens, C. grandiflorum still fits better morphologically with Combretum than with Quisqualis .We suggest it be retained in Combretum sect .Grandiflorae as proposed by Engler & Diels (1899) .Crystalliferous idioblasts in the leaves of C. grandiflorum may not be homologous to those in the leaves of members of sect .Poivrea, but is perhaps a similarity due to convergence .
Nine species in Africa were at one time or another placed under Quisqualis (Liben 1968;Wickens 1973)    species, Q. indica L ., is reputed to be a native of Asia, but is now widely cultivated as a horticultural subject in the tropics .
In Africa, herbarium specimens attributed to Quisqualis indica L .have been recorded from both gardens and the wild in West, East and South Africa .The cultivated Asian form of Q. indica has become semi-naturalized in close proximity to human dwellings, such as in the Nairobi Arboretum (e .g .Williams 396, 529 and Greenway 1821, all in PRE).However, a number of herbarium specimens somewhat resembling Q. indica were collected from natural habitats (near streams or in damp places) along the woodland escarpment in the Iringa District, Tanzania, e .g .Lovett & Congdon 1093, 300, Taylor et al. 8493 (MO, PRE) and Greenway & Kanuri 14816, Napper 1685 (PRE) .These specimens require further study and may well turn out to be an indigenous African form of Q. indica, or a new species.The Tanzanian taxon differs in its petals being well developed, up to 23 mm long with abruptly acuminate apices .Q. indica in Asia has petals up to 12 mm long with round to blunt apices .Wickens (1973) noted the differences between the indigenous Tanzanian form and typical Asian Q. indica, but refrained from taking any formal taxonomic decisions on the status of the African form .In both forms the flowers are white becoming dark red with age, especially on the inside, whereas the leaves have subepidermal, crystalliferous idioblasts, stalked glands, and combretaceous hairs, with hair-tuft domatia in the axils of veins beneath .
Leaves of Quisqualis latialata have black (in herbarium material) stalked glands and combretaceous hairs on both sides, but lack crystalliferous idioblasts; the petals are yellow becoming red with age with a yellow edge; its upper hypanthium is described as cylindric-infundibuliform, very short, less than 6 mm long; the inflorescence is always terminal, contracted into a dense subcapitate raceme .Leaves of Q. littorea also have black stalked glands, combretaceous hairs and hair-tuft domatia in the axils of veins below; the hypanthium is narrowly tubular, up to 20 mm or occasionally up to 25 mm long, petals 6 mm long and apex rounded .Q. pel-legriniana has very thick pubescent young leaves that become puberulous with age and stems with unusual long, flattened, stalked glands mingled with combretaceous hairs, whereas Q. littorea and Q. falcata, both have almost papery, puberulous leaves .The stamens and styles are included or very slightly exserted in Q. falcata, Q. latialata, Q. pellegriniana and Q. littorea, but the first three species have a short (6-10 mm long) upper hypanthium, whereas in Q. littorea the hypanthium can be up to 25 mm long .Liben (1968) lowered the rank of Q .mussaendiflora (Engl .& Diels) Exell to a variety of Q. falcata .Jongkind (1992Jongkind ( , 1999) ) placed Q. pellegriniana and Q. littorea in synonymy under Combretum falcatum (= Quisqualis falcata), resulting in two widely disjunct distribution ranges for the expanded Q. falcata, a classification not accepted by us because there are slight, though quite marked differences between these taxa .In 1991, Jongkind described Combretum inflatum from Gabon, a species with tubular greenish flowers and an adnate style which seems closely related to Q. falcata and Q. latialata .Jongkind (1999) provided a key and illustration (pl .14) to show the differences between C. inflatum, Q. falcata and Q. latialata .They all have greenish flowers, conspicuous floral bracts, whereas the leaves have stalked glands, lacking any scales or idioblasts .The inflorescences of these species are dense, congested capitate racemes with few, long, elliptic to falcate floral bracts, the latter best developed towards the base of the inflorescence axis .The inflorescences also display pronounced acropetal maturation, with the flowers opening sequentially over a period of time from the base to the apex .This phenomenon is so far only displayed by these six species .In essentially all other members of Combretum and Quisqualis in Africa, floral anthesis in any one inflorescence is more or less synchronous .
Quisqualis hensii and Q. exannulata have long (up to 20 mm), protruding styles, the latter twice as long as the stamens in Q. hensii, a state not found in any other species of Quisqualis examined .Q. hensii is also the only member of Quisqualis with long, exserted stamens .Although Q. hensii has the style adnate to the hypanthium for ± 2 mm, stalked glands on the leaves, stems and flower axis, it has flowers with an infundibuliform upper hypanthium, 9-11 mm long .Initially Q. hensii was included in Combretum sect .Campylogyne by Engler & Diels (1899), then transferred to Quisqualis sect .Combretopsis Exell (1931), provisionally associated with Combretum (Exell & Stace 1964) and subsequently placed in a section of its own, namely Combretum sect .Pseudoquisqualis by Exell (1968) and Stace (1981) .Anatomically, Q. hensii has a distinctive type of epidermis, discussed in detail by Stace (1965Stace ( , 1980) ) as well as an unusual indumentum, namely: a mixture of stalked glands; normal combretaceous compartmented hairs; and on young leaves, thin-walled noncompartmented unicellular hairs .The indumentum is very rare in the family (Stace 1965) .Although Q. hensii has some adnation of the style to the hypanthium, it still fits better in Combretum because of its short upper hypanthium and long-exserted stamens and style (Table 1) .

Quisqualis in Asia, Malesia and eastern Africa
The Asian and Malesian species of Quisqualis, namely Q. caudata Craib, Q. conferta (Jack) Exell, Q. parvifolia Exell, Q. sulcata Slooten and Q. indica L .(Exell 1954) and the East and southern African species (Q.indica and Q. parviflora) all have tubular hypanthiums from 10-25 mm long or up to 90 mm long in Q. indica, with ± included or shortly exserted stamens and styles .Q. parviflora and Q. indica have hairs and occasionally stalked glands on the hypanthium, petiole, midrib and lateral veins of the leaves as well as abundant subepidermal crystalliferous idioblasts similar to those in Combretum mossambicense and C. bracteosum .
A resolution on the taxonomic status of Quisqualis From the above overview of Quisqualis in Africa and Asia, it is obvious that there is a great need for further comparative studies to try and resolve the many uncertainties still prevailing .In our judgement, there is still ample morphological evidence to distinguish between Combretum and Quisqualis .The two genera can easily be keyed out in southern and East Africa, Malesia (Exell 1954) and China (Chen & Turland (2007) .Chen & Turland (2007), for example, separate Quisqualis and Combretum on the basis of a combination of characters, namely, 1, length of the hypanthium; 2, degree of stamen exsertion; and 3, adnation of the style to the hypanthium .If the hypanthium is (17-)50-90 mm long, the stamens are not or scarcely exserted from the hypanthium and the style is partly adnate to the hypanthium, then it is a Quisqualis .Combretum on the other hand has the hypanthium usually shorter than 20 mm, the stamens usually well exserted from the hypanthium and the style mostly free or shortly adnate to the inside of the hypanthium .
In the case of the few African species of Combretum (C . hensii, C. grandiflorum, C. ghesquierei) where the style is shortly adnate to the hypanthium, the longexserted stamens and short hypanthiums can still be used to separate these species from Quisqualis.These three species can therefore remain in Combretum, placed in three different sections in subgen .Cacoucia as classified before by Exell (1968) and Stace (1981), Engler & Diels (1899) and Liben (1968), respectively .
We suspect the other six combretaceous species from West Central and East Africa previously placed in Quisqualis (Table 1) may well constitute one or more genera of their own .Quisqualis exannulata from Angola and the DRC has rose-purple flowers, lacks stalked glands and has leaves without abundant crystalliferous idioblasts .It is most unusual in having a geniculate style which is adnate to the hypanthium and long exserted beyond the petals, whereas the stamens are scarcely exserted .The hypanthium is infundibuliform, slightly curved, 25-35 mm long and has a double constriction, one just above the lower hypanthium and another above the portion where the style bends and becomes free from the hypanthium .The flower bracts are large, leafy, ovate, venose, up to 2 / 3 of flower length and present along the whole length of the inflorescence axis, although decreasing somewhat in size towards the apex .It was previously placed in a genus of its own, namely Campylogyne by Hemsley (1897), a classification which we support .Differentiation among these four groups in Quisqualis and Combretum in Africa is presented in Table 1 .
Potential shared characters for Quisqualis s.l.(if one includes Q. hensii and Q. exannulata despite their anomalous floral structure), may include the persistent petiole bases that develop into curved spines, petals minutely to densely hairy or pilose, fruits 5-winged or 5-angled, characters also shared with species of Combretum subgen .Cacoucia sect .Poivrea.However, spiny petioles could have evolved independently in these taxa because all are woody climbers or lianas in forest or forest margins where they need to climb over dense vegetation in search of sunlight .It is strongly recommended that all of the taxa mentioned in the above overview be included in molecular phylogenetic studies .

Molecular phylogeny
Ample leaf and floral morphology (Liben 1968;Exell 1978;Wickens 1973), anatomy (Stace 1965(Stace , 1969(Stace , 1980;;Verhoeven & Van der Schijff 1974;Tilney 2002) and molecular data (Maurin et al . 2010) are available for the southern and tropical African species of Combretum .Unfortunately, only three species of Quisqualis, namely Q. parviflora (Africa), Q. indica (Asia) and Q. caudata (Thailand) have been included in the available molecular studies (Tan et al . 2002;Maurin et al . 2010) .Tan et al . (2002) studied the phylogenetic relationships of subfam .Combretoideae (Combretaceae) based on a limited sampling of only two species from each genus .They concluded that Quisqualis and Combretum are monophyletic sister taxa, but acknowledge that their sampling was insufficient to establish clear generic limits .In subsequent molecular phylogenetic studies based on more comprehensive, though still limited, sampling (Maurin et al . 2010), Quisqualis as a genus is recovered as a clade embedded within Combretum subgen .Cacoucia .As a consequence of these results, the third author supports Jongkind (1999), Stace (2007) and Mabberley (2008) in considering Quisqualis congeneric with Combretum .However, the potentially misleading effects of poor taxon sampling on phylogenetic analyses and their applications should be kept in mind (e .g .Heath et al . 2008) .To increase the accuracy of phylogenetic estimates in Combretaceae it is strongly recommended that as many as possible of the taxa mentioned in Table 1 be sampled in future studies .
In this paper, however, the first and second authors prefer to retain Quisqualis as a separate genus for practical purposes, considering the evolutionary specialization shown by the particular clade .Although there might be intermediate states in the degree of adnation of the style to the upper hypanthium in some species of Quisqualis and Combretum as noted by Exell & Stace (1966) and Jongkind (1991), there are still enough characters to distinguish between these two genera in East and South Africa and they can easily be keyed out .In Quisqualis the upper hypanthium is elongated into a long tube (Figure 1) .In Combretum, on the other hand, the hypanthium is usually a flattened, short, campanulate, infundibuliform or cupuliform limb above the inferior ovary .In C. bracteosum and C. mossambicense the upper hypanthium is distally broadly infundibuliform, whereas proximally it is subglobose, the two parts being separated by a slight constriction .Furthermore, in Quisqualis the stamens are included or very shortly exserted, whereas in Combretum they are always long exserted (Bredenkamp 2000) .The fact that Combretum subgen .Cacoucia is rendered paraphyletic if Quisqualis is maintained as a genus, may be contested by those strictly adhering to Hennigian phylogenetic philosophy, but this is completely acceptable to those prescribing to the more pragmatic evolutionary school of plant classification (Mayr & Bock 2002;Brummitt 2008) .Nevertheless, to cater for the alternative point of view, a new combination and new name are provided for Q. parviflora below .
Pollination and evolution of adaptive traits in the Combretum-Quisqualis clade Pollination syndromes have the capacity to predict the types of pollinators of given species based on observations of certain floral traits .To achieve pollination, flowers have evolved suites of convergent floral traits, usually involving flower shape, colour, scent and type of reward (Fenster et al. 2004) .Stebbins (1970) pointed out that floral diversity has arisen by divergence into different pollination syndromes (evolutionary specialization) .In Combretaceae, major adaptations linked to different pollination specializations are evident (Stace 2007) .There is, for example, the loss of petals as in Terminalia, contrasted with an enlargement of petals in Quisqualis.Differences in pollination strategy are most likely responsible for the floral differences between Combretum and Quisqualis .In the case of Quisqualis, the elongation of the hypanthium might represent a modification for pollination by long-proboscid sphingid moths (Stace 2007) .This shift towards moth pollination might have occurred independently in different Combretaceae .The white-and yellow-flowered species of Combretum with relatively small and reduced petals and with nectaries, suggest pollination by insects with short mouthparts, notably honeybees, whereas those with red flowers and abundant liquid nectar, including the odd-flowered Quisqualis (Campylogene) exannulata are probably birdpollinated (Faegri & Van der Pijl 1971;Stace 2007) .For Combretum, this is confirmed by casual observations on the southern African species .
Combretum displays different floral traits indicative of different pollination syndromes and separate evolutionary diversification, patterns supported by molecular (Maurin et al. 2010) and morphological data .Some evolutionary trends most probably developed independently within the Combretum-Quisqualis clade .In Combretum subgen .Combretum the leaves all have glandular scales, the flowers have long-exserted stamens, short styles, the colour varies from whitish, yellow or green, the upper hypanthium is short, either little developed and flattened (C.imberbe Wawra), campanulate, infundibuliform or cupuliform, all indicative of pollination by insects (Faegri & Van der Pijl 1971) .
All species of Combretum subgen .Cacoucia lack scales on their leaves but have a great diversity in other leaf characters and floral traits .In sect .Poivrea the leaves have abundant, subepidermal, crystalliferous idioblasts, stalked glands and flowers with campanulate upper hypanthiums, exserted stamens, pinkish to bright red .In C .sect .Conniventia the flowers are bright red, with long-exserted stamens and short hypanthiums, suggesting bird pollination .Their leaves lack scales and glands .In sections Megalantherum (C.wattii Exell) and Oxystachya (C.oxystachyum Welw .ex M .A .Lawson), both endemic to the Kaokoveld (northwestern Namibia and southwestern Angola), the flowers are white or greygreen with a pinkish or reddish tinge and a well-developed nectariferous disc; the leaves are densely pubescent with dark brown stalked glands .The stamens and style are exserted beyond the petals and the upper hypanthium is short .
The flowers of Quisqualis in West Central Africa (excluding Q. hensii and Q. exannulata), East and South Africa as well as India, Asia and Malesia are pendent, ± zygomorphic, white, yellowish to greenish turning red with age; have narrowly tubular hypanthiums slightly widening at the tip, petals and calyx lobes bent backwards; no nectar guides; included styles which are straight and adnate for some distance to the hypanthium; and stamens inserted in two rows with versatile anthers, the latter included or very slightly exserted .This syndrome suggests pollination by long-proboscid insects such as sphingid moths (Exell 1954;Faegri & Van der Pijl 1971) .Exell (1954) also noted that the petals in Quisqualis are rather large in the family, much exceeding the calyx lobes and enlarging during anthesis .

Distribution of Quisqualis parviflora
Quisqualis parviflora is endemic to South Africa (Bredenkamp 2000;Jordaan 2003Jordaan , 2006) ) .Van der Schijff & Schoonraad (1971), Carr (1986) and Bredenkamp (2000) give the distribution of Q. parviflora as occurring in Mpumalanga (Mariepskop and Graskop), KwaZulu-Natal and Eastern Cape .Subsequently, all the specimens claimed to be Q.parviflora in Mpumalanga, were shown to be Combretum edwardsii (Verhoeven & Van der Schijff 1975;McCleland 2002) .Sterile material collected from more inland localities in KwaZulu-Natal and previously named Q.parviflora also belongs to C. edwardsii and/or probably an undescribed species of Combretum .Therefore, Q. parviflora only grows with certainty as a woody climber in mainly coastal forest from Ndumu Game Reserve in the north, through most of KwaZulu-Natal to the Eastern Cape (Port St Johns) in the south (Figure 3) .Although no records have been found in Mozambique, it may well occur there, at least in the far south where it borders KwaZulu-Natal (Carr 1986) .

Taxonomy of Quisqualis parviflora
Quisqualis has microscopic stalked glands similar to those in Combretum subgen .Cacoucia, and the scales so characteristic of Combretum subgen .Combretum are absent (Exell & Stace 1966;Wickens 1973) .Flowers are 5-merous and the fruit of Quisqualis are generally 5-winged .The petals are well developed in Quisqualis: white, pink or red in Q. indica (rangoon creeper), a decorative species originally from Asia and commonly cultivated in gardens, and green in the South African Q. parviflora .Specimens seen on the Aluka Library website, http://www .jstor.org/(accessed December 2010), are distinguished by the code e! in the citations below .Gerrard gave the material he collected with McKen at Umhtoti in KwaZulu-Natal the manuscript name, Quisqualis natalensis (now the type of Quisqualis parviflora) .Sonder (1862) changed the epithet 'natalensis' to 'parviflora' and validly published the name Quis-

Alternative taxonomic treatment
If Quisqualis is sunk under Combretum, a new combination and name is required for Q. parviflora .The specific epithet 'parviflorum' cannot be used in Combretum since it was already used by Reichenbach (1825) .Exell (1953) pointed out that Index kewensis incorrectly attributes C. parviflorum to De Candolle (1828), where it was cited as a synonym of C. micranthum Don, a mistake still reflected in The International Plant Name Index (IPNI), accessed January 2011 .The name C. parviflorum was validly published three years earlier by Reichenbach (1825), thus invalidating C. parviflorum Eichler (1867) .The latter name was proposed for a species from Brazil, but because it is a later homonym, this species is now known by the new name C. vernicosum Rusby (1927) .combretum sylvicola O. Maurin, comb . et nom . nov .